Hardware: Familiar but Refined

At first glance, the Pixel Watch 5 looks almost identical to its predecessor. The circular display, the smooth domed glass, and the signature proprietary band attachment all remain. The new satin pyrite finish gives the case a subtle metallic sheen that shifts between gold and bronze depending on the lighting, offering a more premium aesthetic without straying too far from the design language Google established with the original Pixel Watch. The Steph Curry Special Edition adds unique band and watchface options, though the core hardware remains the same.

Inside, the Qualcomm processor has received a generational bump, promising faster app launches and smoother animations. Battery life gets a slight increase, though Google hasn’t specified exact numbers. The watch continues to support dual-frequency GPS, which was introduced last year. There’s also the same array of sensors, including an optical heart rate sensor, SpO2 sensor, altimeter, gyroscope, and accelerometer. Notably, Google has not introduced a new proprietary charger this year, a small victory for anyone who has accumulated a drawer full of previous Pixel Watch charging pucks.

Gemini on the Wrist, Now Offline

The most significant software changes revolve around Gemini, Google’s AI assistant. For the first time, Gemini can operate offline on the Pixel Watch 5 for simple tasks such as starting timers, launching apps, and setting alarms. During early demonstrations, the offline functionality worked as intended, though it was noticeably slower than when connected to the internet. The tradeoff is understandable, as on-device processing requires more time for language understanding and command execution.

Gemini also becomes more proactive. Instead of waiting for explicit commands, the assistant will surface contextual suggestions based on what’s happening around you. For example, if a text message asks about your availability, Gemini may offer to check your Google Calendar directly from the wrist. The watchface itself gains a new icon that displays real-time information for active timers, Uber ETAs, and ongoing workouts, giving you glanceable updates without needing to open an app.

The pinch gesture and wrist-flick controls, previously introduced via a December software update for older Pixel Watches, are now deeply integrated into Wear OS 7. These gestures allow users to answer calls, dismiss notifications, and navigate menus without touching the screen, making the watch more accessible in situations where the other hand is occupied. Combined with the proactive AI suggestions, the Pixel Watch 5 feels more like a personal assistant that anticipates your needs rather than a passive notification hub.

AI-Generated Watchfaces

Google is also bringing generative AI to watchfaces. You can describe a design idea in natural language, and the watch will generate a custom face on the fly. During a demo, a request for “fat cats and an elegant font” produced something that loosely matched the prompt: pleasantly plump, cat-shaped numerals rendered in a refined serif typeface. It’s a gimmicky feature, but it showcases the potential of on-device image generation and gives users a fun way to personalize their wearable. For now, it feels more like a tech demo than a practical tool, but it signals Google’s intent to make AI a core part of the wearable experience.

Breathing Emergency Detection

The most consequential health feature new to the Pixel Watch 5 is called Breathing Emergency Detection. This is an extension of last year’s Loss of Pulse Detection, which was designed to detect when a user’s heart stops beating. With Breathing Emergency Detection, the watch monitors blood oxygen saturation (SpO2) for sudden, severe drops that may indicate the wearer is struggling to breathe. If SpO2 falls to approximately 83 percent or below for an extended period, the watch will trigger a call to emergency services and notify designated emergency contacts.

To reduce false positives, the watch combines SpO2 readings with heart rate, motion, and altitude data. It also disables the feature after the watch determines you’ve been asleep for 30 minutes, since low SpO2 can occur naturally during sleep apnea or side sleeping. Google emphasizes that this feature is not intended for people with chronic respiratory conditions. Instead, it’s designed to flag acute instances where a person may have stopped breathing due to choking, overdose, or other sudden medical emergencies.

At launch, Breathing Emergency Detection will be available on both the Pixel Watch 5 and Pixel Watch 4 in Europe. Google has submitted the feature for FDA clearance in the United States, and its availability will depend on regulatory approval. The feature builds on Google’s broader push into proactive safety tools, leveraging the watch’s sensors to act as a silent guardian in critical moments.

Sleep Breathing Quality Tracking

In addition to emergency detection, the Pixel Watch 5 will track nightly sleep breathing quality and provide monthly summaries. This feature is designed to help users understand how daily habits affect respiratory health over time. By analyzing SpO2 patterns and breathing rate during sleep, the watch can identify trends that may warrant attention. For example, if your sleep breathing quality consistently declines after days of heavy drinking or poor sleep schedules, the monthly report will highlight that correlation.

This feature will also be available on the Fitbit Air, Google’s lightweight fitness tracker released earlier this year. The monthly updates are meant to complement, not replace, professional medical advice. They’re intended to give users a long-term view of their respiratory health, empowering them to make lifestyle changes before problems escalate.

Blood Pressure Trends Without a Cuff

One of the most talked-about new health features is blood pressure trends. Unlike traditional blood pressure monitors that inflate a cuff, the Pixel Watch 5 estimates blood pressure trends using the optical heart rate sensor and motion sensors. Data is fed into an AI model that analyzes the shape and speed of pulse waves traveling under the skin. This model then approximates whether your blood pressure is trending higher or lower over time.

Importantly, this feature does not provide systolic or diastolic measurements. It’s not meant to replace a medical-grade blood pressure monitor. Instead, it functions as a “check engine light” for your cardiovascular health, alerting you to sustained changes that might warrant a visit to a doctor. The feature requires at least a week’s worth of data before it can generate trends, and it will provide monthly updates once active.

Google is not the first company to explore blood pressure insights on a wearable. Apple and Whoop have both dabbled in similar territory, but Google’s approach uses its extensive AI expertise to estimate trends from pulse wave analysis. The feature is scheduled to roll out in September for the Pixel Watch 3 and later models, including the Pixel Watch 5 and Fitbit Air.

Insulin Resistance Trends: A New Frontier

Perhaps the most innovative health feature is insulin resistance trends. This is the first time a major wearable maker has attempted to passively estimate insulin resistance using sensor data. Insulin resistance is a condition in which the body doesn’t effectively metabolize food, causing the pancreas to produce more insulin to manage blood glucose. Over time, this can lead to fatigue, weight gain, and eventually Type 2 diabetes if left untreated.

The Pixel Watch 5 estimates whether your insulin resistance is trending higher or lower month to month based on signals like heart rhythm, sleep quality, and daily movement patterns. It doesn’t provide direct glucose measurements and it’s not a screening tool for diabetes. Instead, it’s meant to flag sustained metabolic stress that could indicate a developing problem.

Google says it validated the algorithm against blood draws and published its findings in a peer-reviewed study in Nature. This lends credibility to the approach, though real-world performance will need to be evaluated. For many people, insulin resistance can remain asymptomatic for years and may not show up on routine metabolic tests. A traditional diagnosis often requires patients to advocate for fasting insulin and glucose blood tests. If the Pixel Watch 5 can accurately identify trends early, it could be a game-changer for preventive health.

This feature is also scheduled to roll out in September for the Pixel Watch 3 and later, as well as the Fitbit Air. Like blood pressure trends, it will require a minimum of a week’s worth of data and will provide monthly trend updates.

Fitness: AI-Corrected GPS and On-Wrist Strength Training

Google is making bold claims about the Pixel Watch 5’s GPS accuracy. For years, users have complained about dubious GPS maps from outdoor activities, especially in dense urban environments where tall buildings block satellite signals. Last year’s introduction of dual-frequency GPS was a step in the right direction, but this year Google is going further.

The Pixel Watch 5 uses satellite and weather station data, along with 3D building models from Google Maps, to correct GPS data in post-processing. The result, according to Google, is more accurate tracking than even dedicated sports watches like the Garmin Fenix 8 or Apple Watch Ultra 3. This is a bold claim, and the real-world performance will need to be tested, but the approach is promising.

Strength training is also getting an AI upgrade. Google’s AI-powered strength training plans, previously available through Google Health Premium on the phone, can now be used entirely from the wrist. You can see your workout plan, log reps, and move to the next exercise without pulling out your phone. This is a natural fit for the Pixel Watch 5, especially given the renewed emphasis on strength training in the GLP-1 era. The on-wrist experience makes it easier to stay in the zone during a workout.

Software, Pricing, and Availability

All of these health and fitness features feed into the Google Health app’s AI coach, which requires a Google Health Premium subscription. The AI coach was already considered one of the best available, though the bar for AI health coaching is still quite low. Google’s approach combines proactive suggestions, trend analysis, and personalized recommendations in a way that feels more cohesive than most competitors. That said, the effectiveness of these features will depend on real-world usage, and Google’s AI coach remains a work in progress.

The Pixel Watch 5 is available for preorder starting today and will arrive on shelves August 20th. The Steph Curry Special Edition follows on September 3rd. The $50 price increase may give some buyers pause, but the expanded software capabilities could justify the premium for those who value AI-driven health insights. Existing Pixel Watch 3 or 4 owners, however, will find little reason to upgrade, as most of the new health software updates will be backward-compatible.

With the Pixel Watch 5, Google is betting that software can be the differentiator in a crowded smartwatch market. The combination of offline Gemini, proactive health tracking, and AI-enhanced fitness features positions the Pixel Watch 5 as a wellness companion rather than just another wearable. It remains to be seen whether these features work as well in real life as they do in demos, but Google’s latest effort is a clear step forward in the evolution of health-focused wearables.

Source: The Verge News

Key facts

• Price: $899 for the Litter-Robot 5 Pro; $599 for the Evo; $699 for the Litter-Robot 4.
• The 5 Pro has a larger drum, LCD screen, and dual cameras, plus an onboard scale and gas sensor.
• Whisker Plus costs $8 per month and is required for AI health insights and video history.
• Cat ID facial recognition repeatedly confused two very different-looking cats.
• WasteID often failed to distinguish between urine and feces.
• The front-facing camera records visits, but the interior camera only offers live view for now.
• Whisker says the system learns over time and has already improved enrollment; some hardware issues required remote fixes.

I tested the 5 Pro alongside the Litter-Robot Evo, both launched late last year. My household includes two cats: Smokey, a large gray and white tabby, and Boone, a smaller sleek tuxedo. They look nothing alike, which made the AI’s confusion even more frustrating.

The basics: still a great pooper-scooper

As an automated litter box, the 5 Pro excels. The cat enters the large globe, does its business, and leaves. A sensor detects the visit, and after a short wait, the drum rotates to sift waste into a sealed bin below. This removes the daily chore of scooping; I only empty the drawer about once a week. The machine is quiet, and the waste drawer contains odors well. The larger globe than the Litter-Robot 4 is a plus for bigger cats, and the LCD screen is far easier to read than the older LED indicators.

There are now four models in the lineup: the Evo, the Litter-Robot 4, the Litter-Robot 5, and the 5 Pro. All use the same basic sifting mechanism. The 5 Pro adds a front-facing camera, an interior camera, an onboard scale, a gas sensor for waste detection, and the company’s Cat ID facial recognition system. The regular Litter-Robot 5 keeps the WasteID tracking but drops the cameras and Cat ID, while the Evo is a smaller, cheaper option for tight spaces.

In terms of pure scooping, the 5 Pro is the best of the group. Its bigger bin, more intuitive controls, and roomier dome make it a better robot than the Litter-Robot 4. The Evo, meanwhile, felt less sturdy, and its waste drawer didn’t always shut cleanly. It also lacks litter-level monitoring, a night light, and compatibility with Whisker’s litter hopper accessory. The Evo is suitable for fewer and smaller cats, which may be fine for some households but is a noticeable step down.

Cat ID: the facial recognition letdown

The 5 Pro’s marquee feature is Cat ID, which uses the front-facing camera to identify which cat is entering the litter box and then logs that visit to the correct profile. Whisker claims it can distinguish up to five cats, but it failed with just two cats that look nothing alike.

After about a week of use, the app walks you through PawPrint setup, the company’s name for facial recognition. It showed me images the camera had captured, and I was supposed to pick the two best for each cat. The images were all fairly poor. Whisker suggests buying a $29 RFID PetTag for cats that look alike, but a $900 machine advertising facial recognition should not require an add-on to tell a gray tabby from a tuxedo.

Whisker says poor lighting and bad initial images can throw the camera off, and when it’s not sure, the system uses other signals, such as the onboard scale, to guess. That is exactly what caused the bizarre weight swap I experienced: the robot mistook my 13.4-pound Smokey for his younger brother and “lost” two pounds overnight. The error persisted, and within a week both cats were registering the same weight. False information can lead to false conclusions, and a robot guessing my cat’s weight is not helpful.

Correcting the mistakes was a lot of work. I had to reset the weights in each cat’s profile, reassign incorrect visits, and fix video logs. Even after all that, the camera still did not consistently recognize either cat. The company says the system is designed to learn and should improve over time, and it has already tweaked the enrollment process to be more selective about which images are used. But a product that requires this much babysitting out of the box is hard to recommend.

WasteID and health tracking: no actionable insights

Whisker’s other AI promise is WasteID, which supposedly recognizes whether the deposit is urine or feces and can even dispatch certain waste faster to reduce odors. It also aims to provide weight measurements for each deposit. In my testing, WasteID repeatedly failed to identify the type of waste, and it often couldn’t provide a weight at all.

Whisker discovered that my unit’s gas sensor, which is meant to sniff out feces, was off and corrected it remotely. When that didn’t improve things, the company determined the onboard scale was also off, which may have contributed to both the WasteID and Cat ID problems. Fixing that would require an entirely new machine.

The $8-per-month Whisker Plus subscription covers the AI health features and video storage. Over my six months with the 5 Pro, it did not deliver a single actionable or useful insight. The newer Patterns feature, which rolled out after launch, at least summarizes each cat’s day, including weight, number of visits, and total time spent in the robot. But it still lacks waste-type details, and the weight data is unreliable. The fact that these improvements arrived months after launch means early adopters were paying for features that did not work.

Cameras and privacy concerns

The cameras are a mixed bag. The front-facing camera catches the cat coming in and records clips, which is useful for checking in remotely. But the interior camera, despite what the marketing implies, only provides a live view; there’s no recording of the actual waste. Whisker says the “poop-view” is coming soon, but that doesn’t help buyers today. If you were hoping to monitor the litter tray in detail, you’ll be disappointed.

I did appreciate the safety features: if the robot detects activity while rotating, it stops and sends an alert. The camera let me check whether a cat was trapped or something was obstructing the globe. However, be careful where you place the robot. It took me a week to realize that the front-facing camera not only captured my cats’ bathroom trips but also anyone using the powder room across the hall. There is a setting to record only when a cat is detected, but cats love to follow people into the bathroom, so that’s not a foolproof privacy guard.

Which Whisker robot should you buy?

If you want a reliable automated litter box and don’t care about AI gimmicks, the Litter-Robot 4 remains a strong choice at $699. It does an excellent job of scooping with no facial recognition or waste identification. The Evo at $599 is tempting for small spaces, but its build quality feels flimsier, and I experienced phantom bonnet detachment errors. Unless space is a genuine constraint, the extra $100 for the Litter-Robot 4 is money well spent.

If you do want WasteID and have large cats or up to five cats, the regular Litter-Robot 5 might be worth a look, but save yourself $100 and skip the Pro. You lose the cameras, but you keep the core scooping and WasteID tracking. Given how poorly the AI features performed in my testing, paying a premium for them seems unjustified.

The Litter-Robot is a fantastic invention, and Whisker has proven it can build excellent hardware. But with the 5 Pro, the company fell into a trap that is all too common in the smart home industry: trusting AI capabilities that work in the lab to behave reliably in the messy, unpredictable real world. Home environments are full of variables — changing light, playful cats, curious toddlers, and random visitors — and AI systems often stumble in ways that can be hard to correct.

It’s possible the 5 Pro will eventually deliver everything Whisker promised at launch. The company is actively updating the software, and the latest Patterns feature shows some progress. But as anyone who follows consumer tech knows, you should never buy a product based on promised future updates. Today, the AI just isn’t ready.

Source: The Verge News

Peak Design built its reputation on camera bags, but the new City Line targets a broader audience: urban commuters, remote workers, and anyone who wants a clean, lightweight bag that doesn’t scream “photographer.” The two bags in the line — the 12-liter City Crescent crossbody and the 15-liter City Backpack — are designed for daily use, with a focus on convenience, durability, and style. They both feature the BagLev hook, which is as clever as it sounds.

What is the BagLev hook?

The BagLev hook is a small, nonslip hook integrated into the bag. When you’re at a cafe, restaurant, or anywhere with a table, chair, or ledge, you can hang the bag off the surface instead of placing it on the floor. The hook is designed to keep the bag stable and off the ground, protecting it from the grime that accumulates on sidewalks, subway floors, and park benches.

On the City Crescent, the BagLev hook rotates, giving you more flexibility in how you hang it. The backpack’s hook, however, is fixed in place, which makes it slightly less versatile. Still, both hooks work as intended, and once you start using them, you’ll wonder why every bag doesn’t have one.

Peak Design City Crescent (12L)

The City Crescent is a crossbody bag that quickly became a favorite for casual outings. It weighs just 530 grams (1.16 pounds), making it one of the lightest bags in its class. The main compartment is surprisingly spacious for a 12-liter bag, easily holding a 14-inch MacBook Pro, though without any padding, you’ll want to use a sleeve if you plan to carry a laptop.

The bag is made from Peak Design’s Mono Shell 660D fabric, which is lightweight, rain-shedding, and easy to clean. The “Moon” colorway, a nearly white fabric, picked up some dirt during testing, but it wiped off easily with a damp cloth. The water-resistant material ensures that your belongings stay dry in light rain, making it a dependable choice for city commuting.

One of the standout features of the City Crescent is its organization. There’s a zippered pocket on the front leading edge that fits a 24-ounce water bottle or a compact umbrella. That same pocket is accessible from inside the main compartment, allowing you to grab your bottle without fully unzipping the bag. The side pockets are equally clever: one zippered pocket points away from your body, perfect for a smartphone, while the opposing side pocket has a self-closing magnetic opening that’s ideal for quick access to glasses or other essentials.

Inside, you’ll find a zippered pocket for a wallet, two stretchy pockets for smaller items like cables and chargers, and a hidden compartment designed to hold a tracker such as an Apple AirTag or a Pixel Tag. The shoulder strap is wide and comfortable, although lacking padding, and it was still comfortable when carrying the bag fully loaded with a laptop, water bottle, umbrella, and assorted accessories.

Peak Design City Backpack (15L)

For those who regularly commute with a laptop, the City Backpack is the better option. It weighs 750 grams (1.6 pounds) and features padded shoulder straps that do a good job of supporting a heavier load. The straps lack pockets or attachment loops, and there’s no sternum strap, but the bag remains stable and comfortable when walking or cycling.

The backpack includes a deep internal laptop compartment that suspends your laptop in the air, with padding on both the front and back. It easily fits a 14-inch MacBook Pro, though larger laptops may not get the same level of protection. There’s also a separate sleeve for a tablet, as well as a variety of internal pockets that offer a mix of zippered and stretchy storage options.

Two stretchy, zippered side pockets are perfect for holding 24-ounce water bottles, but they also work well for carrying a bottle of wine or a baguette — a pleasant surprise for those who like to pick up groceries or snacks on the go. The pockets are accessible by sliding the bag off one shoulder and swinging it forward, a move that grows on you quickly.

The BagLev hook on the backpack is located inside a small pocket at the top of the bag. Unlike the Crescent’s hook, it doesn’t rotate, which can make it trickier to hang from surfaces with short overhangs. The backpack does have an integrated grab handle that can help with hanging, but the limitation is noticeable. Still, it’s a minor issue in an otherwise well-designed bag.

Design and construction

Both bags share the same clean, not-too-slouchy silhouette. They look at home in coffee shops and casual business settings alike. The Mono Shell fabric has a smooth, textured finish that feels durable without being stiff. Large, easy-to-grab zipper pulls make access quick, even with gloves on. There are also thoughtful touches like a key tether and a hidden AirTag pocket.

Peak Design’s reputation for quality is evident in the stitching and materials. The bags feel solid and well-made, and despite their lightweight construction, they don’t feel flimsy. The water-resistant coating adds peace of mind, though the bags are not fully waterproof — they’re designed to shed rain, not survive immersion.

Comparison and pricing

The City Crescent is priced at $119.95, and the City Backpack costs $179.95. Both are competitively priced for the features they offer, especially considering the brand’s focus on thoughtful design. While they lack the protective padding of dedicated camera bags, they appeal to a wider audience who want a versatile everyday carry.

For those who want to try the BagLev hook without buying a new bag, the hook can be removed and attached to other bags that feature similar fabric loops. There are also inexpensive alternatives like portable tabletop hooks from Walmart, Etsy, and Lihit Lab, ranging from $3 to $10. These options provide a similar function, though they don’t integrate as seamlessly as Peak Design’s solution.

Everyday practicality

The City Crescent quickly became the go-to bag for exploring villages and running errands. Its compact size and smart organization make it ideal for carrying essentials like a phone, wallet, glasses, and a water bottle. The BagLev hook meant the bag was never set on the ground, which kept it cleaner and more hygienic.

The City Backpack was the preferred choice for trips that required a laptop. It offers solid protection for a 14-inch MacBook Pro and enough organization for all the tech accessories one might need. The side-access pockets were a particular highlight, making it easy to reach a water bottle without removing the bag.

Both bags excelled in daily use, and the BagLev hook proved to be more than a gimmick. It changes the way you interact with your bag, keeping it off the dirty floor and within easy reach. After a few weeks of testing, it’s clear that the hook is the standout feature, and it makes other bags feel less capable.

The City Line strikes a balance between Peak Design’s lighter Outdoor lineup and its more rugged Travel and Everyday collections. With a focus on weight, durability, and everyday usability, these bags are an excellent choice for urban dwellers and digital nomads.

Peak Design’s City Crescent and City Backpack are thoughtfully designed, lightweight, and practical. The BagLev hook is a genuinely useful innovation that enhances the carrying experience. While neither bag offers dedicated camera protection, they are perfect for everyday use, and the hook alone might be enough to make a purchase.

Source: The Verge News

In an interview, Langley said he has changed his mind on what responsibility Flock bears for how law enforcement uses its technology. The company will roll out updates over the coming weeks that include making Audit Assistance a mandatory feature. This tool flags abnormal search behavior and immediately locks out suspicious users until an administrator reviews their activity. Previously, the feature was optional, and only about a third of agencies working with Flock had chosen to enable it.

The Abuse Reports

The changes come in the wake of a series of disturbing reports about officers allegedly using Flock for stalking and harassment. One police chief in Braselton, Georgia, searched his ex-girlfriend’s license plate more than 500 times over a seven-month period. In California, prosecutors alleged that a former Riverside County Sheriff’s Office deputy used data from Flock to stalk and harass his ex-girlfriend, including looking up license plate information on men she was dating. Reports of Flock and other automatic license plate reader (ALPR) abuse date back to 2021, according to the nonprofit Institute for Justice, and a recent spate of incidents has brought them further into public view.

Langley now predicts that with the new auditing tools in place, more officers will be held accountable. “You are going to see more officers fired and more officers arrested for abusing police power,” he said. He noted that Audit Assistance makes it easier for departments to surface suspicious behavior instead of manually combing through thousands of lines of logs.

Policy Changes

Flock is also shortening the default period it keeps data from devices like its ALPRs from 30 to seven days, moving closer to the limits suggested by critics such as the American Civil Liberties Union (ACLU). Police departments can still choose to lengthen that retention period, or use Flock’s new Evidence Mode, which preserves searches tied to a specific case number. Additionally, individual localities can now set granular limits on data sharing with other agencies. Previously, sharing was all-or-nothing; now a police department can allow another agency to search its Flock data for a homicide suspect while blocking access for immigration enforcement.

All users will also be required to enter case codes to run searches, ideally limiting the ability to run a query for personal or nefarious purposes. This requirement can be overridden by an administrator for emergency situations, such as tracking a missing child, but those overrides will automatically be flagged for review. Langley said the case code requirement was an obvious way to curb abuse, and he admitted, “Fine. You know what? I think that’s right, and we got this one wrong. We should make it a requirement.”

Skepticism from Civil Liberties Groups

After Flock announced the updates, the ACLU released an analysis of the changes, saying they appear “more focused on addressing a perceived PR problem than the significant harms its products create.” While the ACLU called the shorter data retention period a potential “step in the right direction,” it noted that the reality will depend on how Evidence Mode is implemented. The organization also called for independent review of Flock’s auditing and search reason requirement tools to determine how effective they really are.

The Electronic Frontier Foundation (EFF) and other groups have also accused Flock of misleading the public and officials about the risks of its tools, and warned that they could be put to uses like tracking women seeking abortions. In at least one case of allegedly misleading statements, Flock reportedly chalked up the disconnect to confusion over a nuanced issue. The company has separately said officers are prohibited from doing searches related to reproductive care in states where such searches are restricted.

A Shift in Tone

Langley’s public stance has softened considerably. In the past, he took a combative approach to some critics, saying that local policymakers should decide what guardrails are appropriate for its tools. He previously called DeFlock, a site that tracks Flock’s technologies across the country, a “terroristic” group. In July, he apologized and called that characterization a “mistake.”

Langley says his approach has evolved thanks to frequent conversations with communities around the country, including at city council meetings. He understands that many critics will not immediately trust his intentions. “All I guess I’m asking for is for a fair shot,” he said. He still prefers legislators to make key calls about how policing tools should be used, and he hopes that these proposals will also force competing companies to adopt similar guardrails. But he is now publicly acknowledging that regulation could take a while to happen, and that the company bears some responsibility for how its technology is deployed.

“It took me maybe longer than it should have to realize the sheer impact of the company,” Langley said. “I’m still surprised. I got our July update a week ago, and we helped find almost a thousand missing people, over 22,000 stolen cars — in July. That’s crazy. That’s such a big impact on the world. And what I came to realize over time was that the responsibility that bears with it is an obligation to make sure that it’s creating a society that we both, you and I, and everyone around us, want to raise their kids in.”

Flock’s technology has become increasingly widespread in recent years. The company’s ALPRs are mounted on police cruisers and street poles, capturing license plates and using that data to identify vehicles of interest. The systems are marketed as a tool for solving crimes and finding missing people, but privacy advocates have long raised concerns about the potential for abuse. The data collected by Flock can reveal where someone has been, when they were there, and with whom they associated, creating a detailed map of a person’s movements. This information can be valuable for legitimate investigations, but it also presents significant risks when placed in the hands of officers with malicious intent.

The recent incidents of stalking and harassment have highlighted those risks. In the Braselton case, the police chief’s repeated searches of his ex-girlfriend’s license plate were discovered only after an audit was conducted. The former Riverside County deputy’s alleged abuse came to light after prosecutors filed charges against him. These cases demonstrate that without robust auditing and oversight, law enforcement agencies cannot reliably prevent misuse of surveillance data.

Flock’s decision to make Audit Assistance mandatory is a direct response to these failures. The tool is designed to detect patterns that may indicate abuse, such as an unusually high number of searches on a single plate outside of normal work hours. When such behavior is flagged, the system locks the user out until an administrator can review the activity. This type of automated oversight is crucial, as manual reviews are often impractical for departments that process large volumes of data.

The requirement for case codes is another important safeguard. By forcing officers to enter a case number for each search, Flock creates a paper trail that can be used to hold officers accountable. This reduces the likelihood that an officer will run a query for personal reasons, such as checking up on a former partner or tracking someone who angry them. While the system can be overridden in emergencies, those overrides are automatically flagged, ensuring that even legitimate uses are subject to review.

However, civil liberties groups remain concerned about the broader implications of Flock’s technology. The ACLU and EFF have argued that the company has downplayed the risks of its tools and excessive data collection. Even with shorter retention periods, the data collected by Flock can still be used to build detailed profiles of individuals over time. The ability to share data across agencies, even with granular controls, raises questions about how that data will be used beyond the initial purpose for which it was collected.

There is also concern that the new measures could be circumvented. For example, an officer could use an emergency override without a legitimate emergency, or an administrator with malicious intent could approve questionable searches. Independent audits and regular reviews will be essential to ensure that the system is not being abused.

Despite these concerns, Langley’s acknowledgment that the company got it wrong is a notable shift. It reflects a growing recognition among surveillance technology companies that they cannot simply wash their hands of responsibility by pointing to law enforcement policies. As public scrutiny of policing technology intensifies, companies like Flock are being pressured to build more safeguards into their products by design.

The debate over Flock and other ALPRs is part of a larger conversation about the role of surveillance in American society. While the technology can be a powerful tool for public safety, it also raises fundamental questions about privacy and civil liberties. Balancing these competing interests is a challenge that will require ongoing dialogue between technology companies, policymakers, law enforcement, and the public.

Langley’s apology and the new policy changes are a step in that direction, but they do not fully address the systemic risks posed by mass surveillance. The ACLU’s call for independent review of Flock’s tools highlights the need for third-party oversight. Without it, there is a risk that the new measures will be performative rather than effective.

For now, Flock’s changes are set to go into effect in the coming weeks. The company will need to demonstrate that the new safeguards are working as intended and that they are more than a public relations effort. The officers who abused Flock’s tools will face consequences, but the broader question remains: how far should surveillance go, and who is responsible for ensuring it is not used to harm?

Source: The Verge News

An agentic system can be defined as software that turns a user goal into a sequence of steps, executes those steps through tools, keeps track of what happened, and produces an auditable outcome. The model contributes planning and language. The surrounding system provides authority, state, verification, and control. The engineering question remains consistent across domains: how do you give the system enough autonomy to be useful while keeping outcomes predictable? A production answer comes from constraints that are explicit and enforced.

Define the agent loop

An agent loop is the repeated cycle the system follows to complete work. A simple loop should make each stage observable. The plan stage sees the agent choose the next action based on the goal, current state, and policy. The act stage calls a tool with structured arguments and records the result. The verify stage checks the result against policy and task expectations. The commit stage writes the state change to a durable store and produces an audit event.

These words carry specific meanings in implementation. Planning produces a structured intent. Action uses a limited interface with a defined schema. Verification runs deterministic checks. Commit writes the versioned state and the trace context. When each stage is observable, teams can debug failures, understand model behavior, and build confidence for broader autonomy.

Define tools and tool contracts

A tool is any callable capability outside the model. It can be an API, a database query, a workflow engine, a code repository action, or a browser automation step. Tool use dominates operational risk because tools can change systems of record. A tool contract is the boundary that makes tool use safe to operate. It should be written down as part of design review. A contract includes inputs, permissions, idempotency, rate limits, error semantics, and audit fields.

Inputs are defined by a schema that rejects free-form parameters and enforces types. Permissions include identity context, scopes, and data boundaries. Idempotency uses a request key and a replay rule so retries do not create duplicate changes. Rate limits are set per user, per agent, and per tool to protect shared systems. Error semantics provide stable error codes and retry guidance. Audit fields include request ID, actor, time, target record, and before/after references. This contract turns an agent into a regular distributed system client. It becomes testable, debuggable, and something an operations team can own.

Define policy as executable rules

Policy in an agentic system means rules the runtime enforces on every step. Policy should be treated as an executable module. It sits in the request path, it is versioned, and it emits an audit event on decisions. Common policy domains include data access, tool allowlists, approved destinations for writes, required citations for retrieved material, and refusal rules for restricted requests. Policy starts simple and grows based on incident learning.

Executable policy is a meaningful difference from prompt-level instructions. Instructions can be ignored or misunderstood by a model, but code-level policy is deterministic. When policy is versioned, teams can test changes before deployment and roll back quickly if a new rule introduces problems. Every decision the runtime makes can be traced to a specific policy version, which strengthens governance and audits.

Treat state as a first-class component

State is the durable record of what the agent knows and what it has done. State should be kept outside the model, persisted with a clear schema, and versioned per step. At minimum, store the goal, plan steps, tool inputs and outputs, verification results, and final decision. When retrieval is part of the loop, store retrieved sources as well. This state supports replay during incidents and supports evaluation later.

Teams that keep state only in a conversation buffer lose the ability to reason about behavior at scale. A durable state store supports retries, handoffs, and governance reporting. It also enables the investigation of failures after the fact. Without durable state, an agent becomes a black box and operational teams are forced to trust model outputs rather than verify them.

Use verification as a gate on action

Verification is a set of checks that run before a write and after a tool call. Deterministic checks are preferred whenever possible. Model output should be treated as input to be validated. Examples include schema validation, permission checks, reference integrity checks, and constraints on target systems. For content workflows, verification includes citation coverage and checks for restricted data.

Confidence policy is also useful for high-impact actions. The system can require human approval for certain tools or destinations. Approval works best when it is narrowly scoped to a clear action with context and evidence. Verification gates are especially important in enterprise environments where a hallucinated tool call can update a customer record or trigger an external workflow.

Build an evaluation harness around the loop

Evaluation for agents focuses on end-to-end task completion and safety properties. Task success criteria should be defined as observable facts. The ticket exists. The record was updated with correct fields. The pull request passes checks. The change request has the right approvals. Scenario suites should cover routine tasks and edge cases. Run them with fixed seeds where possible and with stable tool mocks. Also run a small set of live tests against a staging environment with realistic data.

Track metrics that connect to operations: task completion rate by scenario, average steps per task, tool error rate, verification failure rate, human approval rate, and mean time to recover when a tool returns partial results. These metrics reveal whether the agent is becoming more effective or simply producing more activity. Without an evaluation harness, teams tend to rely on anecdotal examples, which hide systemic failures.

A practical reference pattern

Production agents are often built with a supervisor pattern. A supervisor owns policy, routing, and state. Specialized workers handle narrow tasks such as retrieval, summarization for a ticket, or a repository action. Workers run with the minimum permissions required for their contract.

A simplified sketch looks like this:

def run_task(goal, user):
    ctx = start_context(goal, user)
    while ctx.open_steps:
        intent = planner.propose_next(ctx)
        intent = policy.enforce_intent(intent, ctx)
        call = tool_router.bind(intent, ctx)
        result = call.execute(idempotency_key=ctx.step_key)
        checks = verifier.run(intent, result, ctx)
        ctx = commit_step(ctx, intent, result, checks)
        if checks.requires_approval:
            ctx = wait_for_approval(ctx)
    return ctx.outcome

This structure keeps authority in the supervisor. It keeps tool permissions narrow. It gives operations teams a single place to enforce policy and observe behavior. The supervisor pattern also scales across domains because workers can be added or replaced without altering the core control loop.

Operational practices that keep agents stable

Several practices help teams run agents safely in production. Start with low-blast-radius workflows. Read-heavy tasks and draft generation build confidence and instrumentation. Ship with a limited tool allowlist. Expand based on measured outcomes and incident learning. Use staged rollouts. Start with internal users, then a small cohort, then broader exposure. Keep tool schemas strict because free-form tool parameters create unpredictable writes. Set budgets to enforce maximum steps per task, maximum tool calls, and a cost ceiling. Maintain runbooks that include rollback, disable switches per tool, and escalation routes to humans.

These practices are not one-time activities. They require continuous attention as the agent expands to new tools and new user groups. A tool that is safe for read-only work may become dangerous once write permissions are added. A schema that works well for one team may need tighter constraints when another team uses it. The practices are meant to be revisited as part of the normal development cycle.

Minimum viable checklist

Teams should have a written definition of the agent loop with traces at each stage. Tool contracts should include schemas, permissions, idempotency, rate limits, and audit fields. A policy module should be versioned and enforced in the request path. A durable state store should keep step-level records for replay and governance reporting. Verification gates should be placed on writes and high-impact actions. An evaluation suite should measure task completion and safety properties. Operational controls should include budgets, staged rollout, and disable switches per tool.

Constraints are key

Agentic systems fit enterprise work because they connect language interfaces to business systems. The systems operate well when autonomy sits inside explicit constraints. Constraints turn agent behavior into something teams can measure, improve, and trust.

Source: InfoWorld News

Adani, the 64-year-old chairman of the Adani Group, was accused in 2024 of taking part in an alleged scheme to bribe Indian officials for lucrative renewable energy projects and to mislead US investors about the payments. Adani and his companies have consistently denied any wrongdoing. In his first public response to the dismissal, Adani said he accepted the decision with humility and deep respect for the judicial process.

The charges were dismissed with prejudice, meaning the government cannot bring the same case again. The ruling brings an end to a criminal matter that had drawn international attention and intensified scrutiny of the Adani Group’s business practices.

Key facts at a glance

• A US judge dismissed bribery and fraud charges against Gautam Adani with prejudice.
• The judge approved a Justice Department request to drop the case but described the decision-making process as “highly unusual”.
• Senior Justice Department official R Trent McCotter had worked with Adani’s lawyers on the dismissal without consulting the prosecutors and investigators who handled the case, the judge found.
• Adani welcomed the ruling and thanked the judicial process.
• The criminal case is separate from other US legal proceedings involving Adani, his nephew and Adani Enterprises.

How the case unravelled

The criminal case against Adani began with allegations that he and other executives paid bribes to Indian officials to secure favourable contracts for solar and wind energy projects. Prosecutors also accused them of hiding the bribery scheme from American banks and investors who were putting money into Adani’s renewable energy ventures.

Adani and his companies denied every aspect of the allegations, describing them as unfounded. Throughout the proceedings Adani never appeared in a US courtroom. His lawyers fought the indictment and eventually succeeded in persuading the Justice Department to review the case.

In May, the Justice Department asked the court to dismiss the charges. It argued that much of the alleged conduct had taken place outside the United States, making the case difficult to prosecute, and that pursuing it no longer aligned with the department’s priorities. The request came shortly after Adani hired a new legal team led by Robert J Giuffra Jr, co-chair of the US law firm Sullivan & Cromwell and a personal lawyer to President Donald Trump. Giuffra had reportedly met Justice Department officials earlier in the year to raise concerns about the case.

Judge questions the dismissal process

Judge Garaufis did not immediately accept the government’s move. In June, he issued an order saying the Justice Department had not provided a sufficient explanation for abandoning the prosecution. He instructed officials to supply more information, and he began examining whether outside pressure or business considerations had influenced the decision.

In his latest order, the judge singled out R Trent McCotter, the principal associate deputy attorney general, for working with Adani’s lawyers on the dismissal. Garaufis said McCotter had appeared to substitute his own judgment for that of the officials directly involved in the investigation. He noted that the decision had apparently been made without meaningful input from the FBI agents and Securities and Exchange Commission investigators who had built the case.

“The irregularities in the decision to dismiss the indictment are concerning,” Garaufis wrote. He said the process was highly unusual and could damage public confidence in the fairness of the justice system.

$10bn investment pledge raises questions

A major point of concern for the judge was an investment pledge made by Adani in November 2024. Just days after Donald Trump won the US presidential election, Adani announced plans to invest $10bn (£7.5bn) in the United States and create 15,000 jobs. The announcement was made in a social media post congratulating Trump on his election victory.

Later reports said Adani’s lawyers had raised the investment during discussions with Justice Department officials about the criminal case. This prompted the judge to question whether the pledge had influenced the decision to drop the charges. Adani’s lawyers said the investment was never offered in exchange for the dismissal of the case. Judge Garaufis ultimately concluded that the investment had not affected the Justice Department’s decision, but he said the public could draw its own conclusions about how such discussions might affect perceptions of equal justice and the rule of law.

Adani’s response

Adani responded to the ruling with a statement on X, formerly Twitter. “Throughout this challenging period, our faith in truth, fairness and the rule of law remained unwavering,” he wrote. He thanked the judicial system and said the decision would allow the Adani Group to move forward with its global investment plans.

The dismissal of the criminal case does not mark the end of Adani’s US legal exposure. In May, Adani agreed to pay $6m to settle civil allegations brought by the Securities and Exchange Commission. His nephew, Sagar Adani, agreed to pay $12m in a separate settlement with the SEC. Both settlements related to investor disclosures and were resolved without admitting or denying wrongdoing.

Separately, Adani Enterprises, the group’s flagship company, agreed to pay $275m to settle potential civil liability over apparent violations of US sanctions on Iran. That settlement also did not involve an admission of wrongdoing. These matters are civil in nature and are separate from the criminal case that has now been dismissed.

Who is Gautam Adani?

Adani is one of India’s richest men and the founder-chairman of the Adani Group, a sprawling conglomerate with interests in energy, ports, airports, mining, cement and other infrastructure businesses. The group has expanded rapidly over the past two decades and now operates beyond India, but it has also faced periodic accusations of corporate misconduct and political favouritism.

In 2023, the Adani Group was the target of a critical report by a US short-seller that accused it of stock manipulation and accounting fraud. The group denied the claims, but the report triggered a sharp sell-off in Adani company stocks and prompted regulatory scrutiny in India. The recent US criminal case added another layer of legal pressure, though the group continued to secure major investment deals and expand into new areas, including green energy and data centres.

Adani is widely seen as close to Prime Minister Narendra Modi. Both men come from the western state of Gujarat, and Adani’s business empire has grown significantly during Modi’s time in power. Adani’s supporters say he is a self-made entrepreneur who has built world-class infrastructure in India. Critics, however, have repeatedly questioned the relationship between his business interests and the ruling administration.

With the criminal charges now dismissed, Adani is expected to focus on the group’s international expansion and its commitment to US investments. The legal proceedings have nevertheless left unresolved questions about how the government reached its decision and what role powerful lawyers and business leaders can play in shaping high-profile prosecutions.

Source: MSN News

The two companies said they will jointly design, develop, and manufacture an indigenous combat engine for India’s Advanced Medium Combat Aircraft (AMCA) program, which aims to produce the country’s homegrown fifth-generation fighter jets. This partnership is not just a commercial agreement; it is a strategic initiative to build a comprehensive aerospace ecosystem in India, aligning with the government's 'Make in India' and 'Atmanirbhar Bharat' (Self-Reliant India) policies. While the companies did not disclose the financial details of the project, the scale and scope suggest a multi-year, high-stakes endeavor.

Strategic Importance of the Partnership

India has long sought to develop its own advanced fighter jet engines, a goal that has remained elusive despite decades of research and development. The AMCA program, led by India’s Aeronautical Development Agency (ADA), is a flagship project under the Indian Air Force (IAF) to create a twin-engine, stealthy, multirole fighter. A critical challenge has been the lack of a powerful, reliable, and indigenous engine that can meet the demanding performance requirements of a fifth-generation aircraft. Currently, India relies on foreign engines, such as the General Electric F404 and F414, for its existing fighter jets like the Tejas. The partnership with Rolls-Royce, one of the world’s leading aero-engine manufacturers, is expected to bridge this technological gap.

The AMCA program has been in development for several years, with the Indian government allocating significant funds to accelerate its progress. In May, the Indian government earmarked 15 billion rupees ($157 million) to bankroll the AMCA program, in an effort to fast-track the development of the next-generation fighter jets for the Indian Air Force. This allocation, while substantial, is only a fraction of the total estimated cost of the program, which is projected to run into billions of dollars over the next decade. The involvement of private-sector giants like Reliance is seen as crucial to bringing in the necessary manufacturing capabilities, supply chain management, and financial resources.

Anant Ambani’s Vision for Aerospace Self-Reliance

Anant Ambani, executive director of Reliance Industries, emphasized the strategic rationale behind the collaboration. “India’s strategic autonomy requires sovereign capability in critical technologies,” he said in a statement. “Our intent with Rolls-Royce is to combine their world-leading expertise in advanced propulsion with Reliance’s technology, manufacturing, scale and execution capabilities to build an indigenous aero-engine ecosystem in India.” This statement underscores the broader ambition of the partnership: to move from mere assembly or licensed production to genuine co-development and ownership of intellectual property. For Reliance, which has traditionally focused on energy, petrochemicals, telecommunications, retail, and financial services, this venture represents a major diversification into high-technology defense manufacturing.

Anant Ambani, the younger son of Mukesh Ambani, has been increasingly taking on leadership roles within Reliance Industries, particularly in new energy and technology ventures. His involvement in this defense project signals a generational shift at the helm of India’s most valuable company. The move also aligns with the Indian government’s push to involve private industry in defense production, breaking the near-monopoly of state-owned enterprises like Hindustan Aeronautics Limited (HAL). By partnering with Rolls-Royce, Reliance can leverage its vast industrial infrastructure, project management expertise, and supply chain network to deliver complex engineering projects at scale.

Rolls-Royce’s Longstanding Presence in India

Rolls-Royce, headquartered in London, has a long and storied history in India, supplying engines for both civil and military aircraft. The company has been a key partner in India’s defense sector for decades, providing engines for the Jaguar, Hawk, and other aircraft in the IAF’s inventory. Tufan Erginbilgiç, CEO of Rolls-Royce, expressed enthusiasm about the collaboration, stating, “Together, with our existing partnerships and capabilities in India, this marks a major milestone towards building a robust, self-reliant aerospace ecosystem in the country.” Rolls-Royce’s expertise in advanced materials, turbine design, and combustion technology is expected to complement Reliance’s manufacturing muscle.

The new partnership will involve setting up a dedicated aerospace gas turbine complex in India, focused on developing advanced power and propulsion technologies. This facility will not only support the AMCA engine program but will also enable the partners to explore wider collaborations across defense, civil aerospace, and new power and propulsion systems. The scope of this facility suggests that the collaboration could extend beyond military jets to include engines for civilian aircraft, ships, and even power generation turbines, reflecting the broader convergence of technologies across sectors.

India’s Quest for Fifth-Generation Fighter Jets

The AMCA program is central to India’s future air superiority. Currently, the IAF operates a mix of Russian, French, and indigenous aircraft, but the modernization of the fleet has been slow due to budget constraints and technical challenges. The induction of fifth-generation fighters, which feature advanced stealth, sensor fusion, and network-centric warfare capabilities, is critical to countering the growing air power of China and Pakistan. India had previously considered purchasing the F-35 from the United States or the Su-57 from Russia, but geopolitical and cost factors led to the decision to develop an indigenous aircraft. The AMCA is designed to be a twin-engine, stealthy aircraft with a payload capacity of over 6.5 tons and a top speed of Mach 1.8.

One of the biggest hurdles in the AMCA program has been the engine. A fifth-generation fighter requires an engine with high thrust-to-weight ratio, supercruise capability (supersonic flight without afterburners), and advanced thermal management to reduce radar signature. Developing such an engine from scratch is a monumental engineering challenge, which even established aerospace powers have struggled with. For India, partnering with Rolls-Royce offers a shortcut to acquiring this critical technology, while also preserving the possibility of future exports. The engines developed under this partnership are expected to power not only the AMCA but also potential future variants and possibly other indigenous aircraft programs.

Implications for the Global Aerospace Industry

The Reliance-Rolls-Royce partnership is a significant development in the global aerospace landscape. It reflects a broader trend of Western defense manufacturers forming strategic alliances with Indian conglomerates, driven by India’s large market and its importance as a geopolitical counterweight to China. For Rolls-Royce, this partnership provides a foothold in one of the world’s fastest-growing defense markets, as well as access to Reliance’s extensive manufacturing and engineering capabilities. For Reliance, it is a bold step into a domain dominated by state-owned enterprises, positioning the company as a leader in India’s defense manufacturing ambitions.

The success of this venture will depend on several factors, including technology transfer, funding, and the ability to navigate complex regulatory environments. India’s defense procurement and industrial policies have often been criticized for bureaucracy and delays, but recent reforms have sought to streamline processes and encourage private participation. The partnership will also need to ensure that the intellectual property generated from the co-development is shared equitably, a sensitive issue in international defense collaborations. However, both companies have expressed confidence in the project, with Anant Ambani stating, “Our intent with Rolls-Royce is to combine their world-leading expertise in advanced propulsion with Reliance’s technology, manufacturing, scale and execution capabilities.”

Mukesh Ambani and Reliance’s Growing Defense Ambitions

Mukesh Ambani, the chairman and largest shareholder of Reliance Industries, is estimated to have a real-time net worth of $89.9 billion, making him the wealthiest tycoon in the Asia-Pacific region. Under his leadership, Reliance has transformed from a traditional energy and petrochemicals conglomerate into a diversified powerhouse with interests in telecommunications (Jio), retail, media, and financial services. The foray into defense aerospace is the latest chapter in this transformation, reflecting a strategic vision to tap into the high-growth potential of India’s defense sector, which is expected to see substantial procurement spending over the next decade.

Reliance has been steadily building its defense portfolio in recent years. The company has entered into partnerships with other global defense majors, including a joint venture with Israel’s Rafael Advanced Defense Systems for missile systems, and has been involved in manufacturing components for naval ships and submarines. The Rolls-Royce partnership, however, is the most ambitious yet, as it involves the development of a complex, high-performance military engine—a technology that only a handful of countries truly master. If successful, it would place India in an elite group of nations capable of designing and producing advanced fighter jet engines, significantly enhancing its strategic autonomy and defense export potential.

The announcement has been met with optimism in industry circles, with analysts noting that it could catalyze the development of a broader aerospace supply chain in India. The dedicated aerospace gas turbine complex is expected to create thousands of skilled jobs and boost indigenous research and development. It will also provide opportunities for small and medium-sized enterprises in India to become suppliers to the aerospace sector, thereby fulfilling the government's vision of an industrial ecosystem that supports self-reliance.

Challenges and the Road Ahead

Despite the promise, the road ahead is fraught with challenges. Developing a new fighter jet engine is a decade-long endeavor, requiring billions of dollars in investment and relentless testing and iteration. The AMCA program itself has faced delays, with the first prototype expected to roll out around 2026 and the first flight projected for 2028-2030, although these timelines may slip. The engine development is on a parallel track, and any setbacks could affect the overall program timeline. Moreover, the integration of a new engine with the aircraft’s airframe, avionics, and stealth features is a complex engineering exercise that demands close collaboration between the engine manufacturer, the aircraft designer (ADA), and the end-user (IAF).

Another challenge is the transfer of technology and export controls. Rolls-Royce, like other Western defense companies, is subject to strict regulations regarding the export of sensitive military technology. While the partnership is intended to build indigenous capabilities, India will need to ensure that it has sufficient autonomy over the critical technologies involved, including the ability to maintain, upgrade, and eventually produce the engine independently. This will require careful negotiation of licensing agreements and intellectual property rights, which are often the most contentious aspects of such collaborations.

Financially, the project will require sustained investment beyond the initial government allocation of 15 billion rupees. Reliance Industries, with its deep pockets, is well-positioned to fund the venture, but the returns are likely to be long-term and subject to the vagaries of defense procurement cycles. The success of the project will also depend on the political will to continue prioritizing indigenous defense industrialization, which has waxed and waned over the years. Nevertheless, the partnership represents a landmark moment in India’s defense evolution, signaling that the country is serious about developing its own critical technologies.

In a statement, Rolls-Royce’s CEO Tufan Erginbilgiç highlighted the broader potential: “Together, with our existing partnerships and capabilities in India, this marks a major milestone towards building a robust, self-reliant aerospace ecosystem in the country.” That ecosystem, if realized, could have far-reaching benefits for India’s economy, its military readiness, and its standing in the global aerospace industry. For now, the announcement serves as a powerful indicator of where India’s defense ambitions are headed, and the role that private industry will play in achieving them.

Source: Forbes News

From Arsenal's continued pursuit of Julian Alvarez and Victor Osimhen to Manchester United's search for a new striker and defensive target, the gossip mill is working overtime. Liverpool have been handed a major boost in their attempts to sign two Paris Saint-Germain attackers, while Chelsea are on the verge of completing a deal for a Spanish left-back. Here is the latest roundup of transfer news, rumours and speculation.

Arsenal step up Julian Alvarez pursuit

Arsenal have reportedly agreed personal terms with Manchester City forward Julian Alvarez as the Gunners look to bolster their attacking options. The Argentina international has been a long-standing target for Mikel Arteta, and the club are now said to have turned their attention back to Alvarez after failing to lure Vinicius Junior from Real Madrid.

According to multiple reports, Alvarez is considered a backup option to Vinicius, but the club are confident of securing a deal. The 24-year-old has grown frustrated with his limited starting opportunities at City, where he often plays second fiddle to Erling Haaland. Alvarez joined City from River Plate in 2022 for an initial fee of £14 million and has since scored 36 goals and provided 18 assists in 103 appearances across all competitions. His versatility, work rate and eye for goal make him an attractive proposition for a side looking to add firepower.

Arsenal's interest in Alvarez comes amid reports that Gabriel Martinelli, Ethan Nwaneri and Gabriel Jesus could all leave the club before the window closes. The Gunners are also said to be keen on signing young forward Christian Kofane, while they have held talks over Napoli striker Victor Osimhen. However, a move for Alvarez appears to be the most advanced, with personal terms reportedly already agreed. The player is said to prefer a move to Barcelona but has not ruled out a return to the Premier League, where he has already adapted to English football.

Manchester United eye Igor Matanovic and Jorge Salinas

Manchester United are exploring options to strengthen their attack and defence before the deadline. The Red Devils are reportedly interested in Eintracht Frankfurt striker Igor Matanovic as a potential addition to their frontline. Matanovic, 22, has impressed in the Bundesliga with his physical presence and clinical finishing, and United are said to be monitoring his availability.

The Croatian forward began his career at Dinamo Zagreb before moving to Germany, where he has developed into one of the league's most promising young strikers. He scored 12 goals and provided four assists in 31 league appearances last season, catching the eye of several top European clubs. United's interest in Matanovic comes as they look to provide competition and cover for Rasmus Hojlund, who arrived at Old Trafford last summer with a big reputation but has yet to consistently find the net.

On the defensive side, United are exploring alternative targets to Lewis Hall. Spanish outlet AS claim that Racing Santander defender Jorge Salinas is being monitored, with Mundo Deportivo reporting that the club are ready to trigger his €16 million (£13.7 million) release clause. Salinas, who can play at left-back or centre-back, impressed during the Under-19 European Championships with Spain. United scouts are said to have been particularly impressed by his ability to read the game and his attacking forays down the flank. The 19-year-old has three years remaining on his contract at Racing and is also attracting interest from Barcelona, which could spark a bidding war.

Liverpool boosted by PSG duo omission

Liverpool have received a significant boost in their pursuit of Paris Saint-Germain forwards Bradley Barcola and Ibrahim Mbaye. Both players were left out of PSG's squad for the Trophee des Champions on Sunday night, fuelling speculation that they could be on the move before the deadline.

The Reds remain eager to secure a blockbuster deal for Barcola, who has been one of the standout wingers in Ligue 1. The 22-year-old France international scored 15 goals and provided 11 assists in all competitions last season, and his direct running and creativity have made him one of the most sought-after young attackers in Europe. However, PSG are holding out for a huge £145 million asking price, and Liverpool will need to negotiate hard to bring him to Anfield.

In addition to Barcola, Liverpool are also interested in signing his PSG team-mate Mbaye. The 18-year-old Senegal international winger has been on the radar of Liverpool's scouting department for some time. Per reports, a move has been explored that would be entirely separate to the pursuit of Barcola. Mbaye could cost up to £43 million, and he has also attracted strong interest from Bayer Leverkusen. His inclusion in the PSG squad for the Trophee des Champions would have been expected had he been fully committed to the club, but his absence suggests that a move could be imminent.

Meanwhile, Liverpool's only signing of the summer so far, winger Victor Munoz, has arrived at the club. Munoz's £34.5 million release clause was triggered in June, but his debut has been delayed by Spain's run to World Cup glory in North America and his subsequent post-tournament break. He has now reported for duty after cutting that holiday short and is confident of making a swift debut in Sunday's pre-season friendly against Monaco at Anfield.

Chelsea finalise Pep Chavarria and remain defiant on Fernandez

Chelsea are expected to finalise the signing of Spanish left-back Pep Chavarria this week. After weeks of negotiations with Rayo Vallecano, the Blues have agreed a deal worth around £16.3 million plus add-ons. Chavarria has already agreed personal terms with the club, and the move is set to be completed in the coming days, all being well.

The 20-year-old full-back has impressed in La Liga with his defensive solidity and overlapping runs, and he is viewed as a long-term investment by Chelsea's recruitment team. He will provide competition for Marc Cucurella and Ben Chilwell in the left-back position.

Despite speculation linking Enzo Fernandez with a move away, Chelsea are remaining defiant. The Argentina midfielder has been the subject of interest from several European clubs, but the Blues are adamant that he is not for sale. Manchester City are also targeting Pedro Neto, but Chelsea are not willing to sanction an exit for the Portuguese winger without a replacement being lined up.

Tottenham step up Gakpo pursuit and agree Phillips exit

Tottenham are stepping up their pursuit of Liverpool forward Cody Gakpo. According to Dutch outlet Voetbal International, new Spurs boss Roberto De Zerbi is pushing the interest aggressively. Gakpo, who joined Liverpool in January 2023 from PSV Eindhoven, is reportedly open to the move, and Liverpool could be tempted to sell for around £70 million.

The 25-year-old Netherlands international has been a key player for Liverpool, scoring 21 goals in 68 Premier League appearances. However, with the arrival of Victor Munoz and the continued development of other attackers, Gakpo's minutes could be reduced this season. De Zerbi is a known admirer of the forward, and a reunion in north London could appeal to all parties.

In other Tottenham news, the club have agreed a permanent deal to send centre-back Ashley Phillips to Middlesbrough. The Championship club will pay an up-front fee of £7 million plus potential future add-ons for the 21-year-old, who is set to undergo a medical. Spurs have also negotiated a sell-on clause in the deal. Middlesbrough have been chasing Phillips all summer and now look set to raid Tottenham for the second time this window after securing the signing of striker Will Lankshear last month in a club-record move that could reach as high as £20 million.

Reijnders keen on Nottingham Forest move

Nottingham Forest are exploring a deal worth £55 million for Manchester City midfielder Tijjani Reijnders. The Netherlands international is open to the move as he looks for more regular first-team football and a prominent role under Oliver Glasner.

Reijnders joined City from AC Milan last summer for £46.5 million and made 47 appearances across all competitions, although many came as a substitute. The 26-year-old is a composed and dynamic midfielder who can operate as a deep-lying playmaker or in a more advanced box-to-box role. Atletico Madrid showed interest earlier in the transfer window but were deterred by City's valuation.

Forest are also pursuing Sporting defender Ousmane Diomande and remain interested in Chelsea forward Liam Delap this summer. Delap, who impressed on loan at Ipswich Town last season, is viewed as a potential addition to Forest's attack, while Diomande would strengthen their centre-back options.

Vinicius Junior signs new Real Madrid contract

Arsenal's hopes of signing Vinicius Junior have been dashed after the Brazilian forward agreed a new contract with Real Madrid. The Athletic reported that the 26-year-old has agreed terms on an improved deal after talks with the Spanish club on Wednesday. An official announcement is expected soon.

Arsenal had identified Vinicius as a leading transfer target and were prepared to make a significant move if he became available. However, Vinicius has spent eight years at the Bernabeu, making 375 appearances, scoring 128 goals and providing 100 assists. He has won three La Liga titles and two Champions League trophies since joining from Flamengo in 2018, and his decision to stay in Madrid ends any speculation about a switch to the Premier League.

Arsenal plot new Ezri Konsa bid

Arsenal remain in pursuit of Aston Villa defender Ezri Konsa as they look to bolster their defensive ranks. The reigning Premier League champions are in need of reinforcements amid injuries to the likes of William Saliba and Jurrien Timber. The defensive concerns were highlighted by their concerning 3-1 pre-season friendly defeat by Real Betis in Dublin.

Versatile England centre-back Konsa has been a target throughout the summer window. The Daily Mail now reports that Aston Villa are braced for a new offer from Arsenal after their previous approach fell some way short of a £60 million valuation. Konsa, who can also play at right-back, has been a consistent performer for Villa and is valued highly by manager Unai Emery. Arsenal will need to significantly increase their offer to have any chance of landing the 26-year-old.

Barcelona agree Rodri deal and other news

In other transfer news, Barcelona have agreed a fee for Rodri, the Spanish midfielder currently at Real Betis. The deal is expected to be completed in the coming days, with Barcelona beating off competition from several other clubs. Rodri, 28, has been a standout performer in La Liga and will add depth to Barcelona's midfield.

West Ham are in talks over Joel Piroe as they look to strengthen their attacking options. The Leeds United forward has been on the radar of several Premier League clubs, and West Ham are hoping to secure his signature before the deadline.

Fulham are also active in the market, chasing Raoul Bellanova from Torino. The Italian winger has impressed in Serie A and would provide width and creativity for the Cottagers.

With the transfer deadline fast approaching, expect more twists and turns as clubs rush to complete their business. The next few days promise to be frantic, with deals for Alvarez, Barcola, Gakpo and others potentially shaping the outcome of the season ahead.

Source: The Standard News

"I feel like I already have a relationship with [Mourinho]," Bellingham said. "It's a dream for me to work with a coach like Mourinho. I've only had one chance to see him, but I'm very excited to see what he's like."

Bellingham, 23, is heading into his fourth season at Madrid after a stunning debut campaign that saw him lift both the LaLiga title and the Champions League trophy. However, the subsequent two seasons brought no major silverware, leaving the club eager for a fresh start under Mourinho, who has returned for a second spell in charge after the team struggled under previous coaches Xabi Alonso and Álvaro Arbeloa.

A Dream Working Relationship

Bellingham's admiration for Mourinho is well founded. The Portuguese coach is widely regarded as one of the greatest tactical minds in football history, having won league titles in Portugal, England, Italy, and Spain, as well as two Champions League trophies with Porto and Inter Milan. His first stint at Real Madrid, from 2010 to 2013, was marked by intense rivalries and a record-breaking 2011-12 season in which Madrid clinched the LaLiga title with 100 points, a league record at the time.

The relationship between Bellingham and Mourinho has already shown signs of mutual respect. Bellingham's versatility and attacking midfield prowess fit neatly into Mourinho's preferred systems, which often rely on a dynamic and disciplined central presence. The Englishman's ability to drive forward, score goals, and press aggressively makes him an ideal candidate for the Portuguese coach's high-intensity demands.

"I'm a bit more mature now, I have more experience," Bellingham said, reflecting on his evolution since joining Madrid. "I have a good feeling about this season with the new signings and the coach, and let's see ... It's an honour and a privilege for me to play for this club. I hope it will be for many more years."

A Turbulent Road Back to the Top

Real Madrid's 2025-26 campaign was overshadowed by inconsistency. The club finished second in LaLiga, eight points behind champions Barcelona, and were eliminated by Bayern Munich in the Champions League quarterfinals. Disappointing results against smaller sides and a lack of tactical clarity led to coaching changes, with Xabi Alonso and later Álvaro Arbeloa failing to steady the ship.

Mourinho's return has injected a sense of optimism among the squad and fans alike. His second coming is expected to bring a return to the pragmatic, counter-attacking style that made Madrid a fearsome European force during his first tenure. He inherits a side that, despite recent struggles, remains packed with world-class talent.

Pre-Season Adjustments

Bellingham was the last Madrid player to report for pre-season training, joining his teammates at Valdebebas on Wednesday after an extended break following the World Cup. He admitted that the early days of training have been demanding but said he is relishing the challenge.

"[I've been] running a lot in the heat, playing a lot of football with my teammates," he said. "It's tough, but it's good as well, to enjoy it, this suffering! I really like this kind of hard work."

The midfielder's late arrival was excused given his participation in the 2026 World Cup, where he delivered a standout performance for England. Bellingham scored seven goals in the tournament, helping the Three Lions reach the semifinals before they were eliminated by eventual champions Argentina. His displays on the biggest stage further cemented his reputation as one of the finest players of his generation.

"The World Cup was a great experience for me and my England teammates," Bellingham said. "But I'm ready to come back and compete. Hopefully, we can win."

LaLiga Schedule and Early Challenges

Madrid's start to the LaLiga season is delayed due to Spain's run to the World Cup final, which has pushed back the opening fixtures. Their first league match is scheduled for August 22 against Espanyol at the RCDE Stadium, preceded by a friendly at Schalke this Sunday. The extended break allowed more time for Mourinho to implement his tactical ideas, but it also means Madrid will face a congested schedule once the season begins.

Bellingham stressed that the team is ready for the challenges ahead. The squad has undergone significant changes in the transfer window, with several key arrivals designed to bolster the squad's depth and quality.

New Signings and a Renewed Squad

Real Madrid have been active in the summer market, bringing in Marc Cucurella, Bernardo Silva, Ibrahima Konaté, Denzel Dumfries, and Carlos Espí. These acquisitions address key areas of concern: Cucurella adds versatility and energy at left-back, Bernardo Silva brings creativity and experience in midfield, Konaté is a powerful and athletic central defender, Dumfries offers attacking thrust from right-back, and Espí is viewed as a promising young talent for the future.

Bellingham believes the new arrivals will help Madrid return to the summit of Spanish and European football. "I have a good feeling about this season," he reiterated. "The mix of experienced players and young energy is exactly what we need."

Bellingham's Evolution

Since moving to Madrid from Borussia Dortmund in the summer of 2023, Bellingham has grown into a leader both on and off the pitch. His debut season was extraordinary, with 23 goals and 12 assists in all competitions, earning him the LaLiga Best Player award. Injuries disrupted his second and third seasons, but he still showcased glimpses of the brilliance that made him one of the most sought-after talents in world football.

Mourinho has a history of maximizing the potential of attacking midfielders, and Bellingham's profile suggests they could form a legendary partnership. The coach's focus on tactical discipline and direct play could unlock even greater production from the Englishman, who thrives when given freedom to roam between the lines and arrive late in the penalty area.

"I feel like I already have a relationship with [Mourinho]," Bellingham said, hinting at an immediate connection. "I've only had one chance to see him, but I'm very excited to see what he's like." The mutual respect is evident, and Madrid supporters will be hoping that chemistry translates into trophies.

With the season fast approaching, Real Madrid are eager to put last year's disappointments behind them. The combination of Mourinho's tactical acumen, Bellingham's world-class talent, and a revamped squad has created a palpable sense of anticipation. For Bellingham, the dream of working with Mourinho is just the beginning of what he hopes will be a remarkable chapter at the Santiago Bernabéu.

Source: MSN News

Produced by Mustard, with additional production from London, the song showcases Rema's ability to experiment with sound while maintaining the melodic charm that has made him one of Afrobeats' brightest stars. The production blends a smooth beat with a catchy hook, creating a track that is perfect for weekend playlists and easy listening. Mustard's influence brings an international, polished feel to the record, while London's additional production work adds depth and texture to the overall sound.

In terms of lyrical content, "Tea" leans into playful and romantic themes. Rema's delivery is effortless, gliding over the instrumental with the kind of charisma that has become his signature. The song evokes a warm and joyful atmosphere, making it an easy addition to any social gathering or solo listening session. It's the kind of record that rewards repeat plays, with subtle production details revealing themselves on each listen.

The Music Video

The official music video for "Tea" is now available on YouTube, offering a visual narrative that complements the song's playful energy. The video features Rema in a variety of settings, each scene crafted to enhance the track's relaxed and flirtatious mood. From colorful backdrops to stylish wardrobe choices, the visual component adds another layer to the song's appeal, demonstrating Rema's continued commitment to delivering striking and memorable visuals with his music.

Fans of Rema have come to expect high-quality music videos, and "Tea" does not disappoint. The attention to detail in the cinematography and editing helps elevate the song beyond just an audio experience. It also serves as a reminder of Rema's status as a visual artist who understands the importance of storytelling in the modern music landscape.

Rema's Journey in Afrobeats

Rema first burst onto the scene as a teenager with his self-titled debut EP in 2019. The project included the smash hit "Dumebi," which quickly became an anthem in Nigeria and beyond. His unique style, often referred to as "Afro-rave," blends elements of Afrobeats, trap, and pop, setting him apart from his peers. Over the years, Rema has continued to evolve, releasing a string of successful singles and collaborations that have solidified his place in the global music industry.

In 2022, Rema achieved a career-defining moment with the release of "Calm Down," a track that later received a remix featuring global pop superstar Selena Gomez. The song became a massive international hit, topping charts in multiple countries and earning Rema unprecedented visibility on the world stage. The success of "Calm Down" was a testament to his crossover appeal and his ability to craft songs that resonate across cultures.

Rema's discography also includes hits like "Iron Man," "Woman," and "Soundgasm," each showcasing different facets of his personality and musical versatility. He has worked with some of the biggest names in the industry, including a notable collaboration with DJ Snake on "Disco Magique" and a standout appearance on the soundtrack for the Marvel film "Black Panther: Wakanda Forever." His contribution to that project further cemented his reputation as a versatile and in-demand artist.

The Production Team Behind "Tea"

One of the most intriguing aspects of "Tea" is the involvement of Mustard, a highly successful producer known for his work in hip-hop and R&B. Mustard, whose real name is Dijon McFarlane, has produced hits for artists like Kendrick Lamar, YG, and Big Sean. His signature sound, characterized by simple yet effective beats and undeniable grooves, brings a distinct flavor to "Tea." This collaboration is a fascinating meeting of styles, blending Mustard's LA-influenced production with Rema's Afrobeats sensibilities, resulting in a track that feels both fresh and accessible.

London, who provides additional production, also deserves credit for shaping the final sound of "Tea." Working in tandem with Mustard, the production team has crafted a sonic backdrop that allows Rema's vocals to shine while maintaining a polished and radio-friendly edge. The result is a song that can appeal to both longtime fans and new listeners, further expanding Rema's reach.

The Significance of "Tea" in Rema's Career

"Tea" arrives at a moment when Rema is enjoying the fruits of his labor. Following the immense success of "Calm Down," expectations for his next moves have been high. Rather than rushing a full album or project, Rema has chosen to release a standalone single, giving fans a taste of what he has been working on without committing to a larger body of work just yet.

This approach allows the song to breathe and find its own audience. In the streaming era, singles have become the primary way artists maintain visibility, and "Tea" is a smart addition to Rema's catalogue. It's a song that feels current but not overtly trendy, staying true to his core sound while incorporating contemporary production touches.

The track also highlights Rema's growth as an artist. His voice sounds more confident, and his delivery is more relaxed compared to his earlier work. This maturity is a natural byproduct of years of experience and global exposure, and it bodes well for whatever he has planned in the future.

Fan Reactions and Social Media Buzz

Leading up to the release, Rema took to social media to tease snippets of "Tea," generating excitement and speculation among his fans. The teasers were met with enthusiastic responses, with fans eagerly counting down the days until the full release. Once the song and video dropped, social media was quickly flooded with positive reactions. Listeners praised the catchy hook, the smooth production, and the visually appealing music video. Several fan edits and dance challenges have already started emerging on platforms like TikTok and Instagram, indicating that "Tea" has the potential to become a viral sensation.

The song's flirtatious and upbeat theme makes it a natural fit for user-generated content, and it wouldn't be surprising to see "Tea" become a recurring soundtrack for lifestyle and relationship content across social media. Rema's interactive approach with his fans has always been a strength, and this release campaign has been no exception.

Will There Be a New Album?

As of now, there has been no confirmed announcement from Rema or his label Mavin Records regarding a new album or extended project connected to "Tea." The single appears to be a standalone release, though fans are hopeful that more music may be on the horizon. Given Rema's track record of delivering quality projects, it's likely that he is diligently working on something significant behind the scenes. For the moment, "Tea" serves as a refreshing single that showcases where he currently is, both sonically and creatively.

Mavin Records, founded by legendary producer Don Jazzy, has been a pivotal force in Rema's career. The label has nurtured his talent and provided him with the platform to reach a global audience. Any future project from Rema would undoubtedly be a major event, but for now, "Tea" is the latest entry in his discography and a welcome addition to the contemporary Afrobeats landscape.

Why "Tea" Stands Out

There are several reasons why "Tea" is a standout release. First, the production quality is undeniable. Mustard's involvement brings a slick, radio-ready sheen to the track, while London's additional production ensures that the beat never feels empty or repetitive. The rhythm is crisp, the bass is present but not overwhelming, and the melodies are instantly memorable.

Second, Rema's vocal performance is among his most engaging to date. He effortlessly switches between singing and quasi-rapping, demonstrating a versatility that keeps the listener engaged from start to finish. The song's structure is also notable, with a hook that is easy to remember and lyrics that are relatable without being overly complex. It's this combination of accessibility and depth that has always been Rema's trademark, and "Tea" is a clear example of his ability to deliver on that promise.

Finally, the official music video is a strong complement to the song. It's visually appealing, well-directed, and captures the playful spirit of "Tea." In an age where music videos are just as important as the songs themselves for establishing an artist's brand, "Tea" delivers on both fronts. It also gives fans a glimpse into Rema's personality, further strengthening the connection between him and his audience.

As the weekend approaches, "Tea" is poised to be a popular add to many playlists. Whether you're a longtime fan of Rema or just discovering his music, the song offers a taste of why he has become one of the most exciting voices in modern music. With its catchy sound, polished production, and engaging visuals, "Tea" is a reminder of Rema's talent and a promising sign of what lies ahead.

Source: BellaNaija News

The election, held across all 30 local government areas of Osun State, saw residents turn out to choose the next governor of the state. By 11:41 p.m. on the same day, INEC had published more than 3,400 results out of the expected 3,763 polling unit results. This represented 92.53 per cent of the total results required for a full picture of the vote.

Key facts at a glance

• INEC uploaded more than 90 per cent of polling unit results to the IReV portal by late Saturday night.
• 3,763 polling unit results were expected, with over 3,400 uploaded by 11:41 p.m.
• Voting began and ended in most areas by 3:30 p.m., followed by sorting and counting.
• Collation of results was ongoing in some local government areas as figures filtered in.
• Observers recorded 323 electoral offences and 38 security incidents in 726 polling units.
• A polling unit result from Ede North showed Governor Ademola Adeleke leading his main rivals.

INEC uploads results to IReV as collation begins

The commission, led by Prof. Joash Amupitan, began releasing polling unit results on the IReV portal after voting concluded in many parts of Osun. The process was still continuing at press time, with officials in several local government areas sorting and counting ballots before uploading results. The commission had earlier assured voters that it would ensure transparency in the release of results, and the uploads reflect that commitment.

A check of the IReV portal showed noticeable activity in several local government areas. Ife North, Irepodun, and Ife East were among the areas where a good number of polling unit results had already been uploaded. The uploads are considered important because they allow the public to view results at the polling unit level before the official collation process is completed.

The IReV portal was introduced by INEC to improve transparency in the electoral process. It provides access to scanned polling unit result sheets, known as EC8A, and is one of the tools used to verify that results at the ward, local government, and state levels are consistent with what was recorded at the polling units. Over the years, the portal has become a major reference point for observers, political parties, and voters during elections in Nigeria.

Osun governorship election and key candidates

The 2026 Osun governorship election is one of the most closely watched off-cycle elections in Nigeria. Governor Ademola Adeleke of the Accord Party is seeking a second term in office after winning the previous governorship election. His main challengers include Bola Oyebamiji of the All Progressives Congress (APC

Source: MSN News

A Peek Into the Kitchen

In his latest Instagram Stories update, Diljit once again gave fans a peek into his kitchen. The actor was filming his cook, who was busy chopping onions when Diljit decided to tease him about one of his most talked-about cooking skills. The short clip captures the easygoing camaraderie between the two, something that has become a beloved part of Diljit's social media presence.

Jokingly asking him, “Dada, dosa bana rahe ho? [Are you making dosa?]” Diljit waited for his response. His cook smiled and shook his head, making it clear that dosa was not on the menu. Those who have followed Diljit's food-related posts over the years would know that he has often said his cook cannot make dosas. This playful exchange has become a recurring theme in their kitchen videos, much to the delight of fans who look forward to these light-hearted moments.

The Wholesome Spread

After the dosa tease, Diljit asked his cook what he was preparing for the day's meal. His cook listed a rather wholesome spread – paneer, chicken, dal, rice and chapati. Diljit then turned the camera towards the gas stove, giving fans a quick look at the dishes being prepared. The visual of simmering curries and freshly made rotis was enough to make anyone hungry, and it highlighted the comforting, home-style food that Diljit clearly enjoys.

This particular meal reflects the kind of balanced Indian thali that many households across the country prepare daily. Paneer offers a vegetarian protein option, while chicken caters to non-vegetarian cravings. Dal, rice and chapati form the staple base, ensuring a satisfying and nutritious meal. The choice of dishes also underscores Diljit's rootedness in Punjabi and North Indian food culture, which he often celebrates through his posts.

A Running Joke About Dosa

The dosa joke has become a signature element of Diljit's foodie content. Over the years, he has repeatedly mentioned that his cook cannot make dosas, and each time he brings it up, it generates laughter and engagement among his followers. This recurring gag adds a personal and relatable touch to his Instagram updates, making fans feel like they are part of an inside joke. It also shows Diljit's ability to find humour in everyday situations, a trait that has endeared him to millions.

Interestingly, the dosa itself is a beloved South Indian staple that has gained immense popularity across India and abroad. By playfully highlighting his cook's inability to make it, Diljit subtly acknowledges the diversity of Indian cuisine while staying true to his own culinary roots. It also sparks conversations in the comments section, with fans sharing their own experiences and favourite dosa variations.

Diljit's Food Adventures: From Toronto to Tokyo

Diljit's foodie escapades are not limited to his home kitchen. Recently, he celebrated his sold-out Toronto concert with a tiramisu and Diet Coke in his signature style. The post, which showed him indulging in the Italian dessert, was a hit among fans who appreciated his candid and unpretentious approach to celebrating milestones. It also demonstrated his willingness to embrace global flavours while maintaining his distinctive personality.

A few days before the latest kitchen update, Diljit Dosanjh gave fans a peek into his food-filled Japan trip with a new Instagram Reel titled "JATT IN JAPAN". The singer-actor explored Tokyo ahead of his Europe tour, sampling local treats and enjoying the city's food scene. He captioned the video, "Before Europe Tour Shopping in Tokyo." One of his stops was 7-Eleven, where he picked up a coffee and showed fans an onigiri, a Japanese rice ball wrapped in seaweed and often filled with ingredients such as tuna, salmon or vegetables. Diljit also stopped by Siva Coffee, where he enjoyed a Biscoff coffee topped with what appeared to be caramel drizzle. The video offered a glimpse into his curiosity and openness when it comes to trying new foods, further cementing his status as a true foodie.

His Love for Desi Food and Global Flavors

Diljit's Aura Tour also got a desi kickstart with a popular dish that he shared on social media. Whether it's a hearty Punjabi meal at home or street food during his travels, he consistently showcases his love for authentic, flavourful food. His posts often feature regional specialities, from buttery parathas to spicy chaat, and he never shies away from expressing his appreciation for a good meal. This genuine enthusiasm resonates with his audience, who see him as one of them – someone who enjoys food without pretence.

Beyond his food content, Diljit's career has been nothing short of remarkable. From his early days as a Punjabi singer to becoming a Bollywood star with critically acclaimed performances in films like "Udta Punjab", "Sooraj Pe Mangal Bhari" and "Jogi", he has consistently pushed boundaries. His music, which blends traditional Punjabi sounds with contemporary beats, has earned him a global fanbase. He has also made history by performing at major international venues, including the Coachella Valley Music and Arts Festival, where he became the first Punjabi artist to perform on the main stage. His worldwide tours, such as the Dil-Luminati Tour, have drawn massive crowds, and his ability to connect with audiences across cultures is a testament to his versatility as an artist.

Diljit's foodie persona is just one facet of his multifaceted public image. He is known for his humility, his sense of humour, and his willingness to engage with fans on a personal level. Whether he's sharing a behind-the-scenes moment from a concert or a simple video of his cook preparing a meal, he brings an authenticity that is rare in the world of celebrity. His Instagram feed is a mix of music, travel, fashion and food, offering a well-rounded view of his life on and off the stage.

The latest kitchen video also highlights the importance of the people who support celebrities behind the scenes. By featuring his cook, whom he affectionately calls "dada", Diljit gives credit to someone who plays an integral role in his daily life. This gesture of respect and affection is not lost on his fans, who appreciate his grounded nature. It also humanises him, reminding everyone that even global superstars rely on home-cooked meals and the comfort of familiar faces.

Food, for Diljit, is clearly more than just sustenance. It is a way to connect with his roots, share his culture and build bonds with his audience. Through his videos and photos, he invites fans into his world, whether he is enjoying a simple dal-chawal at home or exploring exotic dishes in a foreign city. This openness has made him one of the most relatable celebrities on social media, and his foodie posts are eagerly awaited by fans around the world.

In a recent interview, Diljit spoke about how Punjabi cuisine has influenced his life and career. He mentioned that growing up in a small village in Punjab, meals were always a family affair, with everyone sitting together and sharing food. This upbringing instilled in him a deep appreciation for traditional cooking methods and recipes passed down through generations. He often credits his mother and grandmother for teaching him the value of good food, and he tries to incorporate those lessons into his own cooking whenever he gets the chance.

While his cook handles most of the meal preparation, Diljit has occasionally shown off his own culinary skills. From making simple omelettes to attempting more complex dishes, he approaches cooking with the same enthusiasm and energy that he brings to his performances. These videos are a reminder that despite his fame, he enjoys the simple pleasures of life and is not afraid to get his hands dirty in the kitchen.

The response to Diljit's food posts has been overwhelmingly positive, with fans often sharing their own versions of the dishes he features. His influence extends beyond entertainment, as he has inspired many to try new foods and appreciate the diversity of Indian cuisine. Whether he is promoting local eateries during his travels or showcasing home-cooked meals, he uses his platform to celebrate food in all its forms.

As Diljit continues to tour the world and take on new projects, his fans can expect more delightful food content along the way. Every new post offers a glimpse into his ever-evolving culinary journey, from the streets of Tokyo to the comfort of his home kitchen. His ability to find joy in a simple meal is one of the many reasons why he is so beloved by millions, and it is a reminder that sometimes, the most meaningful moments are shared over food.

We simply love Diljit Dosanjh's foodie posts. Don't you?

Source: MSN News

Demis Hassabis, the co-founder and long-time CEO of DeepMind, announced he is stepping back from the day-to-day management of the lab. He will become chair of DeepMind and chief scientist at Alphabet, the parent company of Google. In his new role, Hassabis says he intends to focus on the big-picture challenges that come with the rapid advancement of AI. He will also remain in charge of Isomorphic Labs, the Alphabet subsidiary that uses AI for biomedical and scientific problem solving.

Hassabis helped found DeepMind in 2010, and the company was acquired by Google in 2014. Under his leadership, DeepMind produced a series of landmark research results, including the game-playing AlphaGo system, the protein-folding model AlphaFold, and the architecture that underpins many modern language models. He has also been one of the most prominent voices in the industry predicting the arrival of artificial general intelligence, or AGI, often described as AI capable of human-like understanding and reasoning across a wide range of tasks.

A new leader for DeepMind

The person inheriting daily control of DeepMind is Koray Kavukcuoglu, a veteran of the lab who has served in several research leadership positions. He will take the title of senior vice president of DeepMind, reporting directly to Google CEO Sundar Pichai. Kavukcuoglu has been deeply involved in the engineering and research efforts behind DeepMind's work for years, and his promotion signals continuity rather than a clean break with the past.

In a memo announcing the transition, Hassabis again leaned into the excitement and anxiety surrounding AGI. He said he has been working toward AGI his whole life and now feels it is close at hand. That framing is consistent with the language he has used in public interviews and keynote talks, where he has repeatedly said that advances in large language models are bringing artificial general intelligence closer than many people expect. His new position at Alphabet is designed to give him more room to think about the societal and technical consequences of that transition, while Kavukcuoglu handles the operational work of training and shipping new models.

Senior researchers leave to found Discovery Loop

While Hassabis is remaining inside the Alphabet ecosystem, four prominent Google researchers are leaving the company entirely. Quoc Le, Oriol Vinyals, Jeff Dean, and Sanjay Ghemawat are the co-founders of Discovery Loop, a new startup focused on using AI to automate scientific discovery. The company is structured as a public benefit corporation, meaning it is a for-profit entity intended to have a positive social impact. OpenAI and Anthropic use similar structures, although the practical meaning of that commitment has been debated.

According to reporting from Wired, Google CEO Sundar Pichai made multiple attempts to convince the four researchers to stay, particularly Dean and Ghemawat. He met with them several times before they ultimately decided to move forward with their plans. The fact that Pichai personally intervened highlights the importance of the group. Discovery Loop was able to raise substantial funding largely based on the reputations of its founders, a sign of how much weight these names carry in the AI community.

Jeff Dean is perhaps the most notable of the group. He was Google's 30th employee and has spent decades building the infrastructure that turned Google Search into a global information index. He is widely respected inside Google, to the point that engineers maintain an unofficial list of 'Jeff Dean Facts' on GitHub, filled with humorous and apocryphal claims about his abilities. More recently, Dean and Vinyals played a central role in designing Google's Gemini models, making their departure a significant blow to the company's AI roadmap.

Quoc Le is the co-inventor of several influential neural network techniques, including sequence-to-sequence learning, which became the foundation for many natural language processing systems. Oriol Vinyals, a former DeepMind researcher and Google scientist, was also a key contributor to the transformer architecture and other advances in machine learning. Sanjay Ghemawat is a distinguished engineer and systems researcher who has worked on Google's distributed computing and storage infrastructure. Together, the four bring a combination of systems engineering and deep learning expertise that is hard to replace.

A summer of high-profile exits

The departure of these four researchers is not an isolated event. Earlier in the summer, Google lost Noam Shazeer, a vice president of engineering and co-lead of Gemini. Shazeer was the lead author of the seminal 2017 paper 'Attention Is All You Need,' which introduced the transformer architecture that powers essentially every modern generative AI system. He has since joined OpenAI, a major coup for Google's chief rival. Shazeer had previously left Google in 2021 to found the chatbot startup Character.AI, but returned when Google licensed its technology and brought him back into the fold. That return now appears to have been temporary.

Around the same time, John Jumper, a DeepMind vice president who led the development of AlphaFold, announced he was leaving for Anthropic. AlphaFold is widely considered one of the most important applications of AI in science because it can accurately predict protein structures, accelerating research in biology and medicine. Jumper was quickly followed by senior DeepMind researchers Jonas Adler and Alexander Pritzel, who were reportedly instrumental in the early development of Gemini. Their move to Anthropic also represented a direct transfer of knowledge from Google to its competitor.

These departures have been accompanied by growing concern that Google's recent AI progress has slowed. The company spent much of last year closing the gap with OpenAI after initially being caught flat-footed by ChatGPT and Microsoft's Copilot. At its I/O developer conference this spring, Google said it planned to release Gemini 3.5 Pro in June. Months have passed, however, and the company has only shipped lower-power Flash models, while the more advanced Gemini 3.5 Pro remains absent. OpenAI and Anthropic, meanwhile, have both released new frontier models with improved reasoning and coding abilities, keeping the pressure on Google.

Pressure on Google's AI ambitions

Google CEO Sundar Pichai attempted to reassure employees and observers with a statement emphasizing the company's strengths. He said Google has amazing talent, world-class compute, and products that bring AI to more people than any other company. There is truth to that: Google still operates one of the largest cloud infrastructures in the world, has deep integration of AI into Search, Android, and Workspace, and continues to make substantial research contributions. But the talent drain is a real problem, and the loss of researchers who were directly involved in Gemini's design raises questions about the next generation of models.

DeepMind remains one of the most respected AI research laboratories in the world. The lab has a long tradition of academic publishing and produced some of the most fundamental advances in deep learning. It is also larger and more well funded than most startups, giving it the resources to attract new researchers and continue pushing forward. However, the culture at DeepMind has changed as it has become more product-focused, and some researchers have expressed discomfort with the military and government contracts that Google has pursued in recent years. The increased closeness between Google and the Trump administration has reportedly been another point of friction, and Jeff Dean in particular was said to be among those who pushed back against that closeness.

The broader AI landscape is also more competitive than ever. OpenAI has moved from research lab to product powerhouse, with ChatGPT becoming a household name and its models being integrated into everything from Microsoft Windows to enterprise software. Anthropic has positioned itself as a safety-focused alternative, attracting top talent from DeepMind and elsewhere. Meta has released its Llama models as open weights, and startups like X.ai and Mistral are also competing for attention. In that environment, retaining the people who built Google's Gemini models is essential to maintaining a competitive edge.

Discovery Loop's public benefit corporation structure is interesting because it suggests the four founders are thinking about scientific discovery rather than consumer AI products. They join a crowded field of AI-for-science initiatives. DeepMind's AlphaFold already demonstrated the potential of AI in biology, and other research groups are exploring ways to use large language models for drug discovery, materials science, and automated laboratory experiments. The startup's founders believe their combined expertise can accelerate this trend significantly.

Hassabis's move to Alphabet also reflects a broader shift in the industry as the most senior AI figures move into higher levels of strategic oversight. Many of the defining technical questions are no longer purely about model architecture; they are about compute, data, regulation, and governance. As AGI becomes a more plausible prospect, the people who invented these technologies are increasingly spending their time on safety, alignment, and the societal implications of the systems they helped build. Google will still benefit from Hassabis's advice and research vision, even if it loses some of its most hands-on builders.

The coming months will show whether Google can maintain its position despite the departures. The company still has a deep bench of researchers and engineers who were not part of the recent exodus, and Kavukcuoglu is an experienced leader who knows the inner workings of DeepMind. But the loss of Jeff Dean, Oriol Vinyals, Quoc Le, and Sanjay Ghemawat, plus the earlier exits of Noam Shazeer and John Jumper, means that the people responsible for some of Google's most important AI breakthroughs are now working elsewhere. With Gemini 3.5 Pro delayed and competitors shipping increasingly capable models, the pressure on Google's AI team has never been greater.

Source: Ars Technica News

A long goodbye

The initial plan was to retire Assistant in late 2025, but the service got a brief reprieve as Google shored up some aspects of Gemini. Now that Gemini is ready to take over, Google will begin ending support for the old system on September 4, but the email notes that it may take several weeks for the process to complete. So those stubbornly clinging to Assistant may still be able to use it a bit longer before the option disappears. For those who made the switch already, nothing will change.

This timeline is significant because it shows that Google has finally reached a point where it feels confident that Gemini can handle the core tasks that users depend on daily. The company has been under pressure to catch up with rivals like OpenAI's ChatGPT and Microsoft's Copilot, and integrating Gemini deeply into Android is a key part of that strategy. Assistant, which is based on older natural-language processing models, has become a relic in Google's AI-first vision.

What gets affected?

By ending Assistant on mobile devices, Google is also killing it on myriad other gadgets that connect to your phone for smart features. So smartwatches, headphones, and cars with Android Auto are also being forcibly moved to Gemini. Most of Google's smart home devices have already transitioned to Gemini. This means the impact extends far beyond the phone itself, affecting the entire connected ecosystem that relies on Assistant for voice commands and device control.

Here's a quick rundown of affected platforms:

• Smartphones and tablets running Android
• Wear OS smartwatches
• Wireless earbuds and headphones that use Google Assistant
• Android Auto-compatible vehicles
• Smart speakers and displays that haven't already migrated

What survives — for now

Assistant won't be totally dead after this process is complete. The email notification doesn't mention Android-powered TVs and set-top boxes, nor are vehicles with Google built-in on the list. These platforms will transition to Gemini-only in the future, but Google doesn't have a date for that yet.

You may also get a reprieve if you have an older device that doesn't meet Gemini's minimum specifications or if you live in a region where Gemini is not supported. However, virtually everyone will lose access to Assistant on phones within the next month or two. After that, you'll have to talk to the gen AI-based Assistant or shut off voice control entirely.

This two-tier approach has frustrated some users who feel that Google is pushing an unfinished product. Android TV and Google built-in car owners are left in limbo, not knowing when their devices will be upgraded. Meanwhile, those with older phones are being told that their hardware simply isn't capable of running Gemini, which means they lose Assistant entirely without gaining a replacement.

The rise and fall of Google Assistant

Google Assistant launched in 2016 as a centerpiece of the Pixel smartphone line and was later rolled out to hundreds of millions of Android devices. It was an evolution of Google Now, the predictive cards feature that Google had introduced earlier. Assistant stood out for its natural conversation flow, fast responses, and deep integration with Google services like Calendar, Maps, and Gmail. For many users, it was the default way to set alarms, send texts, play music, and control smart home devices.

At its height, Assistant was available on more than a billion devices. It was widely regarded as the most capable voice assistant on the market, beating Apple's Siri and Amazon's Alexa in numerous head-to-head comparisons. But the rise of large language models changed everything. Generative AI promised more open-ended reasoning and creativity, something that Assistant could never do.

Google began incorporating AI capabilities into Assistant in 2022, with the debut of LaMDA-powered features. But the launch of ChatGPT in November 2022 accelerated Google's timeline. The company quickly shifted resources to Bard, which later became Gemini, and started work on replacing Assistant entirely. By early 2024, Google was openly talking about Assistant's days being numbered.

Gemini: powerful but confounding

Gemini is Google's flagship generative AI model. It can have freeform conversations, break down complex tasks, and call external tools to get things done. That makes it much more flexible than Assistant, which was limited to a fixed set of commands. For example, you can ask Gemini to plan a trip, draft an email, or analyze a photo, and it will respond with context-aware answers that go far beyond simple web searches.

But Gemini's power comes with challenges. The old Assistant was predictable and reliable. When you said "Hey Google, set a timer for five minutes," it did exactly that, no more, no less. Gemini, by contrast, sometimes goes off on a tangent or forgets that it does, in fact, have access to a particular tool. It may misinterpret a simple command because it's trying to reason about the broader context. This can make it frustrating to use for what's supposed to be a quick voice interaction.

Privacy is another concern. Because Gemini is designed to process complex requests, it often sends more data to the cloud than Assistant did. Google says that Gemini on Android is built with privacy safeguards, including on-device data handling for certain tasks. But the shift is a major change for users who prefer their voice interactions to be as limited as possible.

There's also the issue of trust. Assistant had clear, established behavior for many tasks, and users knew exactly what to expect. Gemini can be surprising — sometimes pleasantly, sometimes not. It might suggest an unexpected solution to a problem, but it could also hallucinate facts or misinterpret a request in ways that Assistant never would.

Despite these issues, Google is committed to making Gemini the future of its consumer AI products. The company has been gradually rolling out Gemini to more devices and regions, and the September 4 deadline is a clear sign that it will no longer maintain two separate assistants. For most users, the transition will eventually be painless, but there may be an adjustment period for Assistant holdouts.

When you ask the old Assistant to turn off the lights, you know it will work. When you ask Gemini to do the same, it might reply with a verbose explanation of how smart home technology works, or it might decide to search the web for "turning off lights" before remembering it can control the bulbs.

Google has said it will continue to refine Gemini based on user feedback, and it's likely that many of these quirks will be ironed out over time. But for now, the transition marks one of the most significant changes to Android since the launch of Google Now in 2012. It's the end of an era, and the beginning of a new — perhaps more powerful, but also more uncertain — one.

The email to users is clear: Assistant is shutting down on Android devices starting September 4. Whether that means you'll be forced into the new AI-powered world or you'll cling to the old system until the last possible moment, one thing is certain. Google has decided that the future is Gemini, and there is no turning back.

Source: Ars Technica News

What the Amendment Says

Under the revised terms, Section VI(2)(B) of the company's user agreement is deleted in its entirety and replaced with a new provision. The new section states that, except as expressly provided otherwise in the agreement, users retain ownership of all rights, title, and interests in any content they post, upload, transmit, send, or otherwise make available through the service. This means that posting content on the site does not strip the user of ownership under copyright law.

However, the amendment also grants the company a broad license to use that content. The license is described as royalty-free, perpetual, non-exclusive, unrestricted, and worldwide. It covers a wide range of activities, including copying, reproducing, modifying, editing, cropping, altering, revising, adapting, translating, enhancing, reformatting, remixing, rearranging, resizing, creating derivative works, moving, removing, deleting, erasing, reverse-engineering, storing, caching, aggregating, publishing, posting, displaying, distributing, broadcasting, performing, transmitting, renting, selling, sharing, sublicensing, syndicating, or otherwise providing to others, and using or changing all such content and communications.

Notably, the license is limited to uses on or in connection with the service or the promotion thereof, including commercial purposes. This is a key distinction from prior versions of many online agreements, which sometimes claimed a license for any purpose, including unrelated commercial exploitation. The new wording narrows the scope of what the company can do with user content, tying it directly to the site and its promotional efforts.

Ideas and Suggestions Also Covered

Another important aspect of the amendment is that it applies not only to content such as comments and posts but also to ideas, suggestions, developments, and inventions that users make available. The company may use these for any purpose, on or in connection with the service or its promotion, without compensation or attribution. This is common in user agreements because companies want to avoid legal disputes over unsolicited ideas that might resemble features they are already developing.

The provision also explicitly states that the company may authorize others to do the same. This means third-party partners, syndication services, or promotional vendors could potentially use user content under the same license. For users, this means that a comment or photo posted to the site could appear in promotional materials, advertising campaigns, or partner publications that promote the service.

Why User Agreements Change

User agreements and privacy policies are frequently updated to reflect changes in law, business practices, and technology. Companies often revise these documents to clarify rights that were previously ambiguous or to comply with new regulatory requirements. In this case, the amendment appears designed to make it clear that the company is not claiming ownership of user content, but rather asking for a license to display and promote it.

This distinction between ownership and license is fundamental in copyright law. Ownership of a work belongs to the person who creates it, unless there is a written transfer of copyright. A license, on the other hand, is permission to use the work in certain ways. By stating that users retain ownership but grant a license, the agreement aims to protect the company while respecting the rights of users.

Many online platforms have adopted similar language over the years. Social media networks, forums, and content-sharing sites all need to have clear terms governing how user-submitted material can be displayed, shared, and promoted. Without such provisions, the platforms could not operate effectively or market themselves using user-generated content.

What This Means for Users

For the average user, the practical effect of this amendment is that anything posted to the website may be used by the company to operate and promote the service. This includes the right to modify, crop, resize, or remix content for presentation purposes. It also includes the right to sell or syndicate content to other outlets, although that sale or syndication must be connected to the service or its promotion.

Users who do not want their content used in this way must consider whether they should post at all. The agreement offers no opt-out mechanism. By using the service, users are deemed to agree to these terms. This is standard practice, but it means that individuals should be thoughtful about what they share online.

The company also includes a reminder in the new text: users should make copies of or otherwise back up any content, personal data, or communications they post, upload, transmit, send, or otherwise make available on the service. This is an important note because content can be modified, removed, or deleted by the company as part of the license. For example, a comment could be edited for clarity, or a photo could be cropped to fit a promotional layout.

Background on the Broader Agreement

The amendment is part of a larger user agreement and privacy policy that governs the use of the website. Such agreements typically include sections on acceptable use, intellectual property, disclaimers, limitations of liability, and dispute resolution. Section VI, where the amended provision resides, deals with content and user contributions. The replacement of one subsection illustrates how companies refine their legal documents over time.

It is not unusual for a company to tailor its user agreement for specific websites within a larger portfolio. Each site may have different features, audiences, and content types, and the legal terms may need to reflect that. The amendment explicitly states that it applies only to the technology news website, not to the media company's other properties. This suggests that the company recognized the need for a different set of terms for this particular site.

The timing of the amendment is not explained in the notice. Companies often make changes without fanfare, especially when the changes are clarification rather than a fundamental shift in policy. Users are generally notified via an announcement on the site or through an email, as was done in this case.

Legal Context and Implications

To understand the significance of this amendment, it helps to look at the broader context of digital content licensing. Under the Copyright Act, the author of a work enjoys exclusive rights to reproduce, distribute, perform, display, and create derivative works. Anyone who wants to exercise those rights must have the author's permission. A user agreement can serve as that permission, granting the platform certain rights in exchange for access to the service.

The scope of the license matters. A non-exclusive license means that users can grant the same rights to others. A perpetual license means that the rights continue even if the user stops using the service or deletes their account. A royalty-free license means that the user cannot demand payment for the company's use. These terms are all present in the new section.

One notable addition is "unrestricted." While the license is limited to uses on or in connection with the service or its promotion, the word "unrestricted" suggests that the manner of use within that scope is entirely up to the company. This could include the right to reverse-engineer content, although that is more relevant to data or code than to typical user comments.

The inclusion of "rent, sell, share, sublicense, syndicate" indicates that the company may engage in commercial activities with user content. For example, the website could sell a compilation of user comments to a third party, or syndicate a story featuring user photos. Provided the commercial use is connected to the service or its promotion, the license would cover it.

Another interesting point is that the license applies to "communications" with the company, not just content posted on the service. If a user sends an email to the company or submits a suggestion through a contact form, the company may use that communication under the same terms. This is why the provision references ideas, suggestions, developments, and inventions. Users should be aware that unsolicited submissions become part of the licensed material.

Comparison with Other Platforms

Many popular websites have similar user agreement clauses. For example, social media platforms often require users to grant them a broad, sublicensable license to display and distribute user posts. Some platforms use the license to run advertisements alongside user content, while others use the content to improve their services. The degree of restriction varies from platform to platform.

The new amendment is somewhat more restrictive than some industry standards because it limits the license to the service and its promotion. Many agreements allow companies to use content for any purpose, including unrelated products and services. By tying the license to the site, the company may be aiming to reassure users that their content will not be repurposed in ways that are entirely disconnected from the website.

However, the scope is still broad. "On or in connection with the Service, or the promotion thereof" can encompass a wide range of activities. Promotional campaigns, advertisements, press materials, and even sponsored content might all fall under this language. For users who are particularly protective of their content, the only way to avoid granting this license is to refrain from posting.

Practical Advice for Users

Given this amendment, users should review their own posting habits. If there is a need to retain exclusive control over an image or a written piece, it is safer not to post it on this site. Also, users should consider the fact that the license is perpetual. Even if a user later requests deletion of their account, the license continues with respect to content that was previously posted, unless the agreement says otherwise.

The company advises users to back up their content, which is a sensible recommendation. Because the company can modify, crop, enhance, or even delete content, the only copy that a user can reliably control is the one on their own device. For personal data, users should review the privacy policy to understand how their information is handled, although the amendment appears primarily focused on content rights.

As with any legal document, users who are concerned about the implications should consider consulting with a legal professional. Terms of service are often written in dense legalese, and the practical consequences may not be immediately obvious. A lawyer can explain how the license might apply to specific use cases.

Development of the Amendment

The amendment notice itself is short and does not provide extensive commentary. It is a formal legal text, not an editorial article. This type of communication is common when companies update their terms of service. The notice is meant to inform users of the change, and continued use of the website constitutes acceptance of the new terms.

It is also worth noting that the amendment is dated. In many jurisdictions, companies are required to notify users of material changes to terms of service. The notice may be part of the website's compliance with consumer protection laws. The fact that the amendment is highlighted in a separate announcement indicates that the company wants to draw attention to it, perhaps because it changes the ownership clause.

While the amendment replaces the entire subsection, the rest of the user agreement remains in effect. Users should read the agreement in full to understand the complete set of rights and obligations. The amendment only addresses one specific subsection, but the overall agreement contains many other provisions governing the relationship between the user and the company.

The Broader Trend

In recent years, there has been growing public concern about the rights that users give up when they use online services. Right-to-use terms have become a sticky point for both users and regulators. Some jurisdictions, such as the European Union, have passed laws that restrict how platforms can use user data. In this context, it is understandable that a media company would want to be more explicit about the scope of its content license.

This amendment appears to be a step toward greater clarity. It explicitly acknowledges that users own their content, which is a user-friendly statement. At the same time, it preserves the company's ability to use that content to run and promote its service. This balance is typical of modern terms of service, which try to protect the platform's operations without asserting ownership over user creative works.

Users who want to protect their content can do several things. They can avoid sharing content that they are unwilling to license. They can watermark images or include copyright notices. They can also choose to use the site only for reading rather than participating in comment sections or uploads. Each of these choices reduces the risk of unwanted use of their content.

Key Takeaways

The main takeaway from this amendment is that content ownership remains with the user, but the user grants the company a broad, perpetual, royalty-free license to use that content in connection with the service and its promotion. This includes commercial purposes and extends to ideas and suggestions. The amendment applies only to this specific technology website within the company's portfolio. Users are advised to back up their content and to be mindful of the license they grant when posting.

The update is a reminder that terms of service are living documents, subject to periodic revision. Users who participate in online communities should regularly review these terms to stay informed about how their content can be used. The amendment was announced directly on the website, but not all users will notice such notices. Making an effort to read terms of service, or at least to understand the sections that deal with user content, is a valuable habit in the digital age.

Source: Ars Technica News

The update is one of the clearest snapshots yet of how OpenAI makes money. The company now relies more on business customers than on individual subscribers. That is a significant shift for a company that became famous through a free-to-use consumer chatbot.

Enterprise growth drives $40 billion run rate

The underlying numbers are moving quickly. OpenAI’s annualized run rate has reached $40 billion, roughly double the level of a year ago. July revenue rose 20 percent month over month, while the number of business customers rose 32 percent.

Run rate is an estimate of future revenue based on current performance. It is not the same as audited annual revenue, but it gives investors a way to track momentum. A $40 billion run rate would make OpenAI one of the fastest-growing software companies in history, even before any eventual public listing.

The crossover was expected, just not this soon. Earlier this year, Friar said the consumer and enterprise sides of the business would reach parity by the end of 2026. The new data suggests parity has come much earlier, and the company is now seeing enterprise as the dominant engine.

What the revenue split means

At the start of the year, the split was 60 percent consumer and 40 percent enterprise, according to Friar’s comment. Consumer subscriptions like ChatGPT Plus and Pro still produced strong recurring revenue, but they also exposed OpenAI to consumer churn and the cost of running high-traffic models.

Enterprise contracts, by contrast, are often larger, stickier, and more tied to measurable business outcomes. Companies do not replace their customer service or coding workflows overnight, and once they harden an AI system into production, they are less likely to leave. That is why investors pay close attention to the enterprise mix.

The acceleration may also reflect the spread of ChatGPT to more organizations. The 32 percent jump in business customers suggests the product is no longer limited to early-adopting technology firms.

From tokenmaxxing to cost per unit of intelligence

Friar also described a change in buyer behavior. “Enterprise customers have moved from tokenmaxxing to focusing on cost per unit of intelligence,” she said. In the first wave of generative AI adoption, some companies gave employees open-ended access to AI tools and paid large bills without a clear sense of what was being produced. The term tokenmaxxing captures that behavior: maximizing token usage without regard for return.

That phase is ending. Enterprise buyers now want to tie spending to outcomes. They are asking how many customer tickets are resolved, how much code is produced and reviewed, and whether the output meets an acceptable quality bar. The shift is putting pressure on AI vendors to demonstrate efficiency, not just capability.

OpenAI is answering on price rather than resisting the trend. Friar pointed to recent reductions across the company’s model range. She also said the newest model is 54 percent more efficient on agentic coding tasks. Agentic coding refers to AI systems that can plan, write, test, and iterate on code with little or no human intervention. Gains in efficiency lower the cost of running those tasks and make the economics more attractive.

Advertising is becoming a real business

Another fast-growing area is advertising. OpenAI began testing ads in ChatGPT in February. Six months later, the advertising business is approaching a $1 billion run rate, Friar told shareholders.

That is a notable development for a company that has mostly relied on subscriptions and API usage. Ads inside a chatbot are a relatively new format, and the company is still testing how to balance revenue with a clean user experience. Even so, the early numbers suggest advertising could become a meaningful part of the business.

If the ad business continues to grow, it will put OpenAI in more direct competition with search engines and social platforms that already sell AI-related advertising. It also gives OpenAI a second source of revenue that is not directly tied to model usage limits.

A turbulent week for leadership

The shareholder meeting itself was scheduled before a difficult week for the company. Denise Dresser, the revenue chief, left after eight months in the role. Her departure was announced one day before the meeting. President Greg Brockman thanked Dresser for building the enterprise foundation, according to the attendee. He also praised her replacement, Dali Rajic, who was introduced to OpenAI by Thrive founder Josh Kushner, according to a person familiar with the recruiting process.

Three days before the meeting, longtime executive Brad Lightcap announced he was leaving. Lightcap has been closely associated with OpenAI’s business strategy, and his exit added to the sense of change at the executive level. The company did not offer additional public details about the transition.

Leadership changes come at a delicate time. OpenAI is navigating rapid enterprise growth, new product categories such as advertising, and the complicated process of moving toward a public listing. The fact that the meeting went ahead under these circumstances suggests the company wanted to reassure shareholders that its financial trajectory remains intact.

Chinese open-source models and the cost debate

Shareholders also heard about competition from Chinese open-source models. There has been a broad assumption in the industry that open-source models are cheaper and will force commercial vendors like OpenAI to keep lowering prices. Brockman pushed back against that idea, saying it is a misunderstanding that open source is cheaper.

The argument rests on more than license fees. Open-source models can be downloaded for free, but they still need infrastructure, GPUs, storage, networking, security, and skilled engineers to operate. For many companies, the total cost of ownership can be higher than buying access to a managed service. There are also concerns about governance, compliance, and long-term maintenance.

OpenAI’s price cuts and efficiency improvements could be seen as a response to the same competitive pressure. By making its own models more efficient, the company can lower prices without necessarily sacrificing margin. The question is whether open-source alternatives continue to improve at a pace that narrows the gap.

Quiet on the listing front

On the question of an eventual public listing, executives gave shareholders nothing to work with. OpenAI has a confidential filing with the Securities and Exchange Commission, and executives said they could not discuss timing because of that process.

The confidential filing has been widely interpreted as a step toward an IPO, but the company has not confirmed a formal listing date. With a $40 billion revenue run rate, a growing advertising business, and enterprise demand accelerating, a listing would likely attract intense investor interest. For now, however, the company is keeping details private.

Source: TNW | Artificial-intelligence News

One number to rule them all

Everything in Zoox's safety case reduces to one number: the predicted rate of collision, injury, and fatality events, expressed as miles per event. This is a single estimate that adds together risk from the driving software, from the vehicle itself, and from fleet operations. The result functions as a gate. Before any safety-relevant software release, hardware change, or revision to operating procedure, Zoox updates the safety case and checks that the combined figure still clears its target.

The use of a single aggregate metric is unusual in the autonomous vehicle industry. Some competitors publish separate metrics for disengagements, collisions, and near-misses. Zoox's approach tries to create a single comparable figure that can be weighed against human performance. This makes the safety case easier to evaluate in principle, but also easier to critique if the underlying assumptions are not transparent.

A benchmark built from human data

Zoox anchors its target in human driving data. The company builds its benchmark from the National Highway Traffic Safety Administration's crash sampling and fatality reporting systems, as well as two Federal Highway Administration datasets. It parses those datasets by road speed and weights them to match the mix of roads its robotaxis actually use. This weighting is important because a fleet operating mostly on low-speed urban streets cannot be compared fairly with a national average that includes highways and rural roads.

By matching the benchmark to its own operating domain, Zoox aims to make the comparison more meaningful. If its robotaxis drive in dense city environments, the human crash rate on those same types of roads becomes the relevant baseline. The company says its target is anchored in this human data, but it does not publish the numerical benchmark itself. That omission makes independent verification difficult.

What remains unpublished

Even though the methodology is described in detail, Zoox does not publish the answer. It sets its target by comparison to the human benchmark and says it aims to be significantly safer than a human driver, but it never defines how much safer counts as significant. It also does not give the figure that its fleet currently reaches. This leaves a critical gap: the public can see how the number is calculated, but not what the number is.

This lack of specificity could be intentional. Safety data in the autonomous vehicle industry is often treated as a competitive advantage, and liability concerns may discourage companies from publishing exact figures. But for regulators and the public, the absence of a concrete number makes it harder to assess whether Zoox is actually meeting its own safety goals. The framework is essentially a promise to calculate risk rigorously, without a clear way to verify the outcome.

The engineering behind the estimate

The engineering described in the safety case is specific enough to argue with. Zoox names the hazard analyses it runs, follows the ISO 26262 functional safety standard with integrity ratings on the platform, and uses simulation that deliberately searches for the conditions where a collision is most likely. Those simulation results are later weighted by real fleet exposure to produce a realistic risk estimate.

This combination of directed simulation and real-world weighting is more sophisticated than a simple log of miles driven. It allows Zoox to explore rare edge cases that may not appear often in normal testing. The company also says it uses separate integrity ratings on different parts of the platform, which means critical systems are held to a higher standard than non-critical ones. This aligns with established automotive safety practice.

A second pair of eyes

Some aspects of the Zoox system are genuinely unusual. The company says it runs a separate collision checker with its own perception system. This independent checker can veto a trajectory that the main planning system has proposed. This is a more robust approach than relying on a single perception stack to both see and act. If one perception system fails to detect an obstacle, the other may still catch it.

Zoox also concedes that a likelihood-based metric cannot capture rare avoidance scenarios. A statistical model may show that a particular situation is unlikely to occur, but if it does occur, the consequences could be severe. To address this, Zoox keeps a test set where the robotaxi must at least match a competent human driver. This test set is separate from the statistical risk model and serves as a safety net for edge cases that probability alone might miss.

Humans in the loop

Another notable feature of the Zoox framework is the inclusion of remote staff inside the model rather than outside it. The company employs TeleGuidance tacticians who never drive. They offer route guidance and high-level suggestions while the vehicle itself remains responsible for all driving decisions. The risk of these tacticians making a mistake, or their tools failing, is priced into the same safety estimate.

This is a different philosophy from companies that use remote operators as a fallback with full driving authority. Zoox argues that keeping human responsibilities limited reduces the cognitive burden on remote staff and makes their behavior more predictable. But the framework explicitly accounts for human error, acknowledging that even a limited role carries some risk.

A framework with teeth

The safety case also reserves the right to restrict, pause, or ground the fleet. This is not a theoretical power. Three weeks before publishing the framework, Zoox recalled all 105 of its robotaxis after one failed to detect heavy smoke and drove into an active fire scene in Las Vegas. That incident involved a vehicle continuing into an area where first responders were operating, a serious failure that could have endangered emergency workers.

The recall was the fourth software recall for Zoox in roughly 13 months. It followed regulatory pressure from the National Highway Traffic Safety Administration, which had demanded fixes for vehicles interfering with first responders. NHTSA had logged 123 collisions involving Zoox vehicles in autonomous mode as of March. Although many of those collisions may have been minor, the volume underscores the challenges of operating a robotaxi service in real-world conditions.

The repeated recalls raise questions about the maturity of Zoox's software. The company has been developing autonomous vehicles since 2014, but its commercial deployment is still early. Each recall is a sign that the safety case is working as intended, catching problems before they cause more severe accidents, but the frequency also suggests that the system is still evolving significantly.

The mileage gap

The analytical approach exists because the mileage does not. Zoox has around three million autonomous miles, a substantial figure for a young company but far behind its main competitor. Waymo passed a hundred million autonomous miles more than a year ago. This gap matters for both engineering and public trust. More miles generally mean more exposure to rare events and more data to validate safety claims.

Zoox now has a paid service to protect while it closes that gap. The company launched its robotaxi service in Las Vegas in 2023, after years of testing in San Francisco and other cities. It offers fares to the public, which means its safety decisions directly affect paying passengers. The publication of the safety framework can be seen as an effort to build confidence among riders and regulators, even as the company acknowledges that its most important metrics remain secret.

For now, Zoox is asking the public to accept a detailed methodology without a final score. That may be a reasonable request in a complex engineering field, but it will likely face continued scrutiny from regulators who have already forced multiple recalls. The path to proving safety is long, and Zoox is still writing it, one miles-per-event calculation at a time.

Source: TNW | Amazon News

Energy Transfer's Transwestern unit told federal regulators on Friday that the Green Chile Project now has an in-service date of 1 February 2027, against a previous target of 15 August. The filing with the Federal Energy Regulatory Commission marks the latest setback for Project Jupiter, a massive data center campus in Doña Ana County, New Mexico, that is expected to run on natural-gas-fueled Bloom Energy fuel cells and host AI infrastructure for OpenAI.

A data center on an unprecedented scale

Project Jupiter is not a typical data center development. It is planned at 2.5 gigawatts, with developers Stack and BorderPlex Digital Assets discussing investments of up to $165 billion across 1,400 acres and four buildings. The scale is so large that the natural gas pipeline intended to serve it would carry up to 400 million cubic feet per day, roughly 0.4% of all gas produced in the lower 48 states.

The numbers are difficult to overstate. A single gigawatt is enough to power around 750,000 homes, depending on the region. At 2.5 gigawatts, Project Jupiter would be among the largest data center campuses ever proposed. The involvement of Oracle as the tenant, with OpenAI as the underlying AI workload, suggests the facility is being designed for the next generation of large-scale machine learning models, which require enormous amounts of electricity and cooling.

The project is also part of a larger shift in how cloud companies are building capacity. Rather than relying solely on centralized cloud regions, hyperscalers and AI companies are increasingly demanding dedicated, highly dense data centers located close to power sources. This has created a new kind of competition, not just for chips and customers, but for land, water, and natural gas pipelines capable of delivering large volumes of fuel on demand.

An elected official in the way

Standing in the path of the pipeline is New Mexico land commissioner Stephanie Garcia Richard, an elected official who oversees state trust lands. She has refused to allow the pipeline to cross state trust land twice, first in March and again on 15 July. Her objections are not procedural. In her written decision, she argued that the project would benefit investors and developers but bring no significant benefits for state lands. She described the burden on New Mexico's water and natural resources as extreme.

On emissions, she was even more direct, saying that approving the pipeline would be “literally doubling down on dangerous emissions” at a time when the state should be slowing climate change rather than accelerating it. The commissioner's stance reflects a growing tension between the data center industry's demand for reliable power and the climate goals of the states where those data centers are built.

Water concerns in a dry state

Water is a particularly sensitive issue in New Mexico, which has been dealing with drought and the shrinking Rio Grande for decades. Data centers themselves use water for cooling, and gas extraction can also place pressure on local water supplies. Garcia Richard's objection highlights the fact that even though fuel cells are more efficient than traditional combustion turbines, the entire supply chain from gas well to power generation has environmental consequences.

The commissioner wrote that the project's impacts on water and natural resources would be profound, and that the economic benefits promised by the developers did not justify the long-term costs to the state. This is part of a broader pattern across the American West, where data center proposals are colliding with water scarcity, wildfire risk, and community concerns about industrial development.

Oracle argues time is of the essence

Oracle has made no secret of its urgency. In a May filing with the Federal Energy Regulatory Commission, the company said that “time is of the essence” and that any delay would jeopardize the broader objectives of Project Jupiter. The company has been borrowing heavily to build data centers as it races to challenge cloud leaders like Amazon Web Services, Microsoft Azure, and Google Cloud in the AI era.

Oracle's aggressive expansion strategy depends on having power in place when customers like OpenAI need it. A delay of nearly six months could ripple through the company's contracts and its ability to deliver AI services at scale. Investors reacted to the news on Friday by sending Oracle shares down about 4%, while Energy Transfer rose 1.4%, a sign that the market sees the delay as a negative for Oracle but not necessarily for the pipeline operator.

The fuel-cell compromise

Interestingly, the fuel cells at the heart of Project Jupiter were already a compromise. Bloom Energy's solid oxide fuel cells replaced the gas turbines and diesel generators that were originally planned. Oracle has since expanded its Bloom order to 2.8 gigawatts, suggesting that it sees the technology as a way to reduce emissions while still maintaining reliable, on-site power. But fuel cells still burn natural gas, and the industry as a whole continues to rely heavily on fossil fuels despite its public promises of sustainability.

Tech companies often market themselves as leaders in renewable energy, but their data centers are increasingly turning to natural gas to bridge the gap between grid capacity and the explosive growth in AI computing. Some companies have signed power purchase agreements with solar and wind farms, but the intermittency of renewables makes them insufficient for facilities that need to operate around the clock. Natural gas, while cleaner than coal, still produces carbon dioxide and methane emissions.

The choice of fuel cells also reflects a broader tension within the industry. On one hand, tech giants have pledged to reach net-zero emissions and are spending billions on clean energy projects. On the other hand, the immediate power demands of AI are so enormous that natural gas has become the most pragmatic option for many hyperscale projects. The Project Jupiter dispute brings this contradiction into sharp focus, forcing public officials to weigh private investment against public goods.

A wider pattern of local resistance

The Project Jupiter delay is also a story about land, not just capital. Money and computer chips can be acquired at speed, but a 17-mile right of way cannot. The pipeline route crosses state trust land, and the land commissioner has the power to veto it. In the broader data center industry, this is becoming an increasingly common obstacle. According to community groups and industry observers, around 500 towns in the United States have already blocked data center development in their jurisdictions, either through moratoriums, zoning restrictions, or outright opposition.

The reasons for local resistance vary. In some cases, it is about noise and visual impact. In others, it is about water use, electrical grid strain, or the environmental footprint of natural gas pipelines. There is also a growing concern that data centers create relatively few permanent jobs after construction is complete, while consuming a large share of local power and water resources. This has led to a backlash in places as different as Northern Virginia, Oregon, and New Mexico.

State trust lands are a particularly tricky hurdle. In New Mexico, these lands were set aside by Congress to generate revenue for public schools and other institutions. The land commissioner is required to manage them in a way that benefits the trust beneficiaries, not necessarily to maximize private commercial returns. That means a project must show a clear and direct benefit to public schools, universities, and other state institutions. Garcia Richard determined that the Green Chile pipeline did not meet that standard, even though the data center itself could eventually generate tax revenue.

The delay is also a reminder of the physical limits of the AI boom. Every large language model, every autonomous vehicle system, and every generative AI application depends on data centers that require unprecedented amounts of electricity. The companies building that infrastructure are discovering that securing permits for power generation and pipelines is often more difficult than building the facilities themselves. Decades-old laws governing land use, environmental review, and public utilities are now becoming critical bottlenecks for the world's most valuable technology companies.

The Green Chile pipeline is now scheduled to go into service on 1 February 2027, nearly six months later than originally planned. The delay may be only the beginning. The land commissioner has rejected the route twice, and it is unclear whether Energy Transfer and the developers can find an alternative path or persuade her to change her mind. The state's trust land board, which oversees the commissioner's office, could also become involved.

For Oracle, the stakes are enormous. The company has staked its future on being able to build hyperscale data centers quickly and cost-effectively. But the physical reality of land, water, and local politics is proving harder to overcome than the technical challenges of chip supply or server deployment. The delay of the Green Chile pipeline is a reminder that in the age of AI, the most valuable commodity is not compute but consent.

Source: TNW | Artificial-intelligence News

• Canva's valuation was cut by $7.1 billion (US) or A$10 billion by backers Blackbird and Airtree, dropping from $42 billion to $34.9 billion, a 17% decline.
• Canva's own internal valuation fell more sharply, from $38.9 billion to $31 billion over the past year — a $7.9 billion reduction.
• The company cut its expected growth rate by a third to 20% after second-quarter revenue missed internal guidance, reaching $921.9 million.
• CEO Melanie Perkins said demand for AI features exceeded expectations, but the cost of completing each AI task was too high, prompting a slowdown in rollout.
• Cost per AI task has fallen nearly 90% since April, while users now create three times as many designs.

The $7.1bn markdown

Canva's longest-standing backers have reduced the design platform's valuation by $7.1 billion in US dollar terms, a cut that reflects the mounting cost of artificial intelligence. The Australian Financial Review first reported the markdown on 14 August, noting that the figure is also A$10 billion, the number used in most headlines. Blackbird and Airtree, two of Canva's earliest investors, now value the company at $34.9 billion, down 17% from $42 billion.

But the more consequential number came from inside the company. Canva's own independent valuation, used to determine the price at which employees can sell their shares, has fallen from $38.9 billion to $31 billion over the past year. That is a $7.9 billion decline, larger than the investors' cut. Two separate valuers, moving in the same direction, paint a consistent picture: the era of easy zero-marginal-cost software is over.

AI bill shock

The markdown arrived a week after Canva's revenue troubles became public. On 3 August, the company admitted it had cut its expected growth rate by a third, to 20%, after discovering that its AI tools were far more expensive to run than anticipated. Second-quarter revenue rose 25.2% to $921.9 million, but that still missed internal guidance. The headline from that earlier report said it directly: Canva had an AI bill shock.

Melanie Perkins, Canva's co-founder and CEO, explained the problem candidly. In an email to Fortune, she said the launch of the company's AI features had “validated the demand, but also showed us we needed to reduce the cost of completing an AI task to support a broad rollout.” The decision followed: “Rather than broadly rolling out a product before the underlying economics were ready, we decided to slow the rollout while we rebuilt the architecture, reduced unit costs and strengthened the business model.”

Since Canva AI 2.0 launched in April, the cost per task has fallen by nearly 90%, according to Perkins. Yet users are creating three times as many designs, so the savings are not as large as the percentage suggests. Net compute spending still rises, but the company is buying itself time to fix the underlying economics before scaling up.

Software's broken secret sauce

The structural issue goes beyond Canva. Derek Hernandez, a senior research analyst at Pitchbook covering SaaS and AI, told Fortune that AI is dismantling a core assumption of the software industry: zero marginal cost. “AI is making SaaS no longer a zero marginal cost solution, which has really been what I would call a lot of software's secret sauce up until now,” he said.

His analogy is instructive. Building a Ford F-150 is the equivalent of training a model; the petrol and the mechanic are inference, “because that's the point of using the product.” In the old software model, serving one more user of a design tool cost almost nothing. Every AI-assisted image now carries a compute bill attached. That changes the unit economics of every transaction, and companies must either raise prices, cut costs, or slow growth.

Five days apart: Canva and Figma

Hernandez also connected Canva's problem to a broader market event. “Canva and Figma both hit the same wall about five days apart, but they cited it in different places,” he said. Figma, which went public earlier this year, grew 48% and raised its outlook, yet its stock fell 16% on margin concerns.

Figma's free cash flow margin dropped to 14% in the second quarter from 27% in the first quarter, and the company guided third-quarter growth to 36%, down from 48%. One company is public and got repriced in a day. The other is private and got repriced by its shareholders. Same wall, same collision.

The same bill is arriving everywhere

It is not just design software. Large cloud buyers are hitting the same cost wall. Amazon reportedly ran a Claude job that went 860% over budget before failing. Microsoft has placed spending limits on internal AI use. EY built an AI router that sends tasks to cheaper models rather than defaulting to the frontier ones, a solution close to what Perkins describes rebuilding at Canva.

Rory O'Driscoll of Scale Venture Partners framed the timing on the 20VC podcast: “There's going to be a lot of people paying the bill in 26 and 27 for a certain amount of hesitancy in 23 and 24.” The hesitancy refers to the period when companies raced to adopt AI without understanding the economics. The bill is now arriving.

Why Canva can't sit it out

For Canva, AI is not a side feature; it is the expansion plan. The company has been pushing beyond its core design tool into enterprise workflows, adding products like Canva Code and aiming to become a one-stop platform for business creativity. Perkins told Fortune in 2023 that the AI market was too fragmented, and the strategy since has been to bring those pieces inside one platform.

That creates a dilemma shared across the sector. Companies cannot afford to skip the AI boom, but embracing it can undermine the margins they are trying to protect. Europe has a new entrant on the same premise — Lovable, which reached $500 million of revenue with just 146 employees, though it has no comparable margin history. The pressure applies equally to newcomers and incumbents.

The listing is the reason for the brakes

Canva was valued at $42 billion in an employee share sale earlier this year and was expected to list in 2026. Hernandez now suggests the listing could come as early as next year. He reads the slowdown as investor-facing: “I'm sure they're trying to protect their profitability, especially if they want to go to public investors.”

Growth of 20% with repaired margins is a very different pitch to growth of 30% with a cost problem. Figma's experience suggests public markets punish the second one immediately. Canva's private shareholders are effectively making the same judgment before the IPO.

What would settle it

Three things will determine whether Canva's correction is a temporary setback or a new normal. The first is the next revenue print. Canva has guided to roughly 20% growth, and whether that holds will show if the cost problem is fixed or merely deferred. The second is the 90% cost reduction claim. A cost per task falling that far while usage triples is testable over a couple of quarters — if the margins improve, the architecture rebuild worked. The third is who else gets marked down for this reason. Hernandez calls Canva and Figma the biggest signals so far, which implies more are coming.

For now, Canva's backers have accepted a lower valuation, the company has accepted a slower growth rate, and the entire software industry is waking up to a new reality: artificial intelligence is powerful, but it is not cheap.

Source: TNW | Business News

Hauser sat down with CNBC's The Tech Download podcast, and the full conversation ran to 40 minutes as a video episode. The discussion spanned AI, semiconductors, quantum computing, and European sovereignty. The written version of that conversation, published on 14 August, distills his arguments into a series of pointed observations. Rather than simply calling AI a bubble or dismissing the skeptics, Hauser separates the technology question from the pricing question. He argues that AI will create more value than probably any other technology revolution ever seen, but that some valuations have clearly gotten ahead of themselves.

Not a bubble call, and not a defense either

Hauser's starting point is far from the typical tech skeptic. He describes AI as a revolution that will generate enormous economic value, perhaps more than any prior technological shift. Yet he immediately adds a qualifier: the journey will be a rollercoaster. That single word captures his view that while the long-term trajectory is upward, the path will be marked by sharp drops, overcorrections, and painful resets.

On the subject of prices, Hauser is specific rather than sweeping. He does not claim that the entire AI sector is overvalued. Instead, he points to structural risks that could trigger a correction. His primary concern is a recent wave of circular financing deals, where chipmakers take equity stakes in AI labs that then use the proceeds to buy chips from those same investors. This self-reinforcing loop can inflate revenue figures and create the appearance of robust demand, but it also concentrates risk in a way that could amplify a downturn.

Why circular financing matters

The Bank for International Settlements flagged this pattern in June, warning that an AI bust could hit credit markets as hard as the 2008 financial crisis. The mechanism is straightforward: if AI labs are buying chips with money that comes from chipmakers, the revenue is not entirely exogenous. A slowdown in AI investment could cause a cascade of losses that ripple through both the tech sector and the financial system. Hauser is not predicting that outcome, but he argues that the largest players are well positioned to survive a reset. OpenAI and Anthropic, in his view, hold significant capital reserves and should withstand turbulence even if expectations are recalibrated around them.

That leaves an obvious question that Hauser does not directly answer. If the leading AI labs survive a reset, someone else absorbs the pain. The circular deals are the most likely point of impact. Chipmakers that have invested heavily in AI startups could see the value of those stakes decline or the expected purchase orders evaporate. Smaller players without the same capital buffers would be more vulnerable. Hauser's position, therefore, is narrow and unusual: the revolution is real, prices are not all real, and the biggest names are not the fragile part of the system.

Set against other tech leaders' views

His comments contrast sharply with Masayoshi Son, the chief executive of SoftBank Group. Son told shareholders in June that calling AI a bubble is an insult, and he has described such commentary as blasphemy. Hauser occupies a more nuanced middle ground. He agrees that the underlying technology is transformative but insists that the market's pricing of specific companies can become disconnected from fundamentals. This dual stance is harder to compress into a headline but offers more practical guidance for investors and policymakers trying to separate signal from noise.

The distinction between the technology question and the pricing question is central to Hauser's thinking. He does not believe that AI is overhyped in terms of its potential impact. He does believe that some companies have been valued as if that potential has already been realized, which creates room for disappointment. This is not a contrarian position for its own sake; it is a disciplined attempt to evaluate the industry on multiple timescales.

The chip architecture argument is the freshest part

Hauser co-created the architecture that powers most of the world's mobile devices, so his views on the next wave of chip design carry particular weight. He argues that AI is forcing a fundamental rethink of computer architecture itself. The pressures are practical and urgent. AI models are expensive to run, chips generate enormous amounts of heat, memory costs are rising, and bottlenecks have appeared throughout the industry. These constraints are not merely engineering problems; they are economic and environmental limits that will shape the next decade of computing.

Hauser points to two responses that aim to address these pressures: in-memory computing and photonic computing. Both approaches try to reduce the energy and time spent moving data between processors and memory. In traditional computer systems, data shuttles back and forth between the processor and memory, a journey that consumes power and creates latency. In-memory computing moves the arithmetic into the memory cells themselves, eliminating much of that travel. Photonic computing uses light instead of electrical signals to transmit data, offering the possibility of much higher bandwidth and lower energy consumption.

The significance of these changes, in Hauser's words, could match the architectural breakthroughs that allowed Arm to challenge the established chipmakers decades ago. Arm succeeded by focusing on power efficiency at a time when the industry was optimizing for raw speed. The constraint of battery life created an opening that Arm exploited, and the same kind of constraint-driven opportunity is now emerging in the AI era. Hauser admitted that he never thought AI would produce such a fundamental shift in computer architecture, yet that is precisely what he sees happening.

Money is already moving toward photonic computing

The photonics half of this argument is not theoretical. Nvidia, the dominant supplier of AI chips, has spent $6.5 billion across photonics companies in just three months. The goal is to replace copper wires with light inside AI data centers, where the sheer volume of data being moved between thousands of chips has become a critical bottleneck. This is a case of a constraint being bought out rather than engineered around. The problem that Hauser describes already has a price attached to it, and major industry players are placing large bets on the solution.

In-memory computing is further behind in terms of commercial deployment, but the concept is gaining traction. Instead of moving data to a processor and waiting for the result to come back, the arithmetic is performed where the data is stored. This approach has the potential to dramatically reduce the energy cost of AI inference and training, which has become a growing concern as model sizes expand. Both photonic and in-memory computing are attempts to address the same underlying target: the cost of moving data, rather than the cost of the mathematical operations themselves.

Arm's history provides a useful precedent. The company won on power efficiency at a moment when the industry was focused on raw performance. The constraint of mobile device battery life created the opening that allowed Arm to become the dominant architecture in phones. Hauser sees a similar dynamic at play in AI. The constraints of energy consumption, heat dissipation, and memory bandwidth are reshaping the industry, and the companies that can solve those problems will be the next major winners.

Europe can build, but cannot scale

When the conversation turns to Europe's position in the global tech landscape, Hauser is direct in both directions. He believes that European companies have the innovation and skill to compete with the United States and China in deep technology. The failure, he argues, comes at the next stage. European startups struggle to grow from small, innovative firms into genuinely global competitors. This is not a new diagnosis, but coming from an investor who has spent years funding that attempted jump, it carries particular weight. Amadeus Capital Partners exists to help European companies make that leap, so Hauser's frustration reflects personal experience as well as structural analysis.

He connects this scaling problem to a larger concern that he keeps returning to throughout the interview: technological sovereignty. Hauser worries that Europe depends on foreign suppliers for critical technologies, ranging from AI models to semiconductor design software. That second item is often overlooked. Chip design software is a small market with only a handful of suppliers, but it sits upstream of everything else in the semiconductor industry. Without access to that software, it is nearly impossible to design advanced chips, regardless of how innovative a European company might be.

The recent moves in that sector demonstrate the risk. Synopsys, one of the leading electronic design automation companies, pulled back from chip fabrication software in July to focus on higher-margin AI design tools. This shift means that even the limited options available for chip design may become more constrained over time. Hauser's warning is that dependency on a few suppliers, especially those outside Europe, leaves the continent vulnerable in an era of geopolitical tension and export controls. Export controls turn a simple supplier relationship into a chokepoint. A tool that can be purchased today is not just a purchase; it is a dependency that can be weaponized tomorrow.

Hauser's caveat is as firm as his warning. He believes that close cooperation with allies is essential, and he is not arguing for a break with the United States. But he insists that partnership and dependence are different arrangements, and Europe has been treating them as the same. The distinction matters because dependence carries risks that partnership does not, especially when trade policies, export restrictions, and political tensions can shift quickly.

The line that captures a broader fear

His conclusion on this topic was blunt and quotable. Europe should preserve its partnership with the U.S., but it should not become a technology colony of the U.S. That phrase encapsulates the fear that Europe's technological future is being shaped by decisions made in Washington and Silicon Valley rather than in European capitals. Brussels has already begun to act on that fear. The European Commission proposed a sovereignty package in June that would restrict U.S. cloud providers from handling sensitive government data, alongside a Chips Act 2.0 aimed at boosting domestic semiconductor production. These are steps in the direction Hauser advocates, but he would likely argue that they are not enough.

His position is not one of simple anti-Americanism. He acknowledges the importance of the transatlantic alliance and the value of cooperation on technology and security. But he draws a clear line between partnership, which is a relationship between equals, and dependence, which leaves one side vulnerable to the other's decisions. Europe, in his view, has been treating these two arrangements as interchangeable, and that is a dangerous mistake.

The broader lesson from Hauser's interview is that the AI revolution must be evaluated on multiple dimensions. The technology itself is transformative and will create enormous value, but the market around it is subject to the same cycles of overvaluation and correction that have hit every previous technology wave. The physical infrastructure supporting AI is changing in fundamental ways, driven by energy and bandwidth constraints. And the geopolitical landscape is being reshaped by the concentration of critical technologies in a few players. These are not separate issues; they are interconnected forces that will determine who benefits from the AI era and who is left behind.

What would settle the questions Hauser raises? First, the architecture claim is testable. In-memory and photonic computing are already being developed and deployed, and Nvidia's investment in photonics shows that the market is treating these approaches seriously. Over the next few years, it will become clear whether these technologies can deliver the dramatic efficiency gains that their proponents promise. Second, the question of whether a European company can make the jump from startup to global competitor is an open one. That is a test of growth capital, management talent, and market access, not just research capability. Third, the rollercoaster that Hauser expects will inevitably arrive, and the circular financing deals he flagged are where the evidence of trouble will show up first. If a reset occurs, the resilience of the largest AI labs and the vulnerability of smaller players will become apparent, and the market will have a clearer picture of who was building real value and who was merely participating in the hype.

Source: TNW | Artificial-intelligence News