Meta's advertising platform has become the focus of another child safety scandal. An investigation revealed that the company approved and ran dozens of paid advertisements containing explicit AI-generated child sexual abuse imagery over a nine-month period. Some of these ads remained accessible even after Meta was directly confronted about them. The findings add to a long list of child protection failures that have followed the social media giant for years.

Ads Hidden in Meta's Own Transparency Tool

Researchers at the Tech Transparency Project discovered the offending ads inside Meta's public Ad Library, a tool the company created to provide advertising transparency. The ads were live between November 2025 and early August 2026 and reached users in the United States, the United Kingdom, and more than a dozen European countries. At least one advertisement was delivered to over 2,500 accounts in Europe alone.

The violations were not subtle. According to the Tech Transparency Project's director, the images made no attempt to hide what they were promoting. These were not pieces of user-generated content that slipped past moderators. They were paid advertisements that Meta reviewed and approved before collecting ad revenue from them.

Several ads linked directly to so-called nudifying applications, including one called MaskAI. Apple removed that app from its App Store after the researchers flagged it to the company. In a particularly concerning development, the research team initially found around two dozen violating ads. But in the hours before the report was published, they discovered roughly 30 more. Several of those new ads had gone live only after Meta had already been asked about the problem.

The investigation also highlighted a troubling pattern: some of the ads matched content that Meta had previously taken down for exactly the same type of violation. That raises serious questions about whether Meta's enforcement systems are capable of recognizing repeat offenders. Many of the accounts running these ads had very few followers. Where advertiser information was visible, several accounts traced back to Chinese firms, including one previously identified as a former Meta ad reseller.

Meta responded with familiar language. The company called the content horrific, noted that most ads had limited reach, and pointed to a new AI detection system. It also cited a statistic claiming that 36 million pieces of child exploitation content were removed last year. However, critics say these large numbers do little to comfort users when illegal ads are still able to pass through the review process and reach thousands of accounts.

The latest investigation also underscores how much the online advertising industry has changed. AI-generated content can now be produced quickly and cheaply, allowing bad actors to test platforms at scale. Ad libraries, which are meant to increase transparency, have become treasure troves for researchers looking for violations. But the same public nature of these ad archives also gives malicious advertisers a roadmap for what can slip through.

A Long History of Child Safety Failures

This is not the first time Meta has faced public scrutiny over child safety. The company's problems date back to at least 2017, when Facebook was already receiving more than 50,000 reports of revenge porn and sexual abuse every month. At the time, Meta hired thousands of content moderators to keep up with the flood of reports.

In 2019, internal messages later revealed in court showed that Meta changed how Messenger shared information with law enforcement. That change affected roughly 7.5 million child abuse reports, making it harder for authorities to access critical evidence.

In 2021, whistleblower Frances Haugen leaked internal research showing that Meta's own studies found Instagram was harming teenage girls' mental health. That same year, engineer Arturo Bejar emailed Meta's CEO directly about his own daughter's experience with sexual harassment on Instagram. Despite those warnings, the platform continued to operate largely as before.

By 2023, a joint investigation by the Wall Street Journal, Stanford University, and UMass Amherst found that Instagram's algorithms were actively connecting accounts linked to pedophilia. The investigation described a network of child sexual abuse material traders who used Instagram's recommendation systems to find each other. Bejar later testified before the Senate that Meta's safety features were nothing more than a placebo in name only. Senators called the evidence damning and promised to take action, but no binding regulation has been passed, and Meta's practices have remained largely unchanged.

A Landmark Lawsuit and a $375 Million Verdict

The warnings eventually moved from congressional hearings to the courtroom. New Mexico's Attorney General sued Meta after an undercover investigation found that a fake profile of a 13-year-old girl was immediately flooded with sexual solicitations from adults. The case went to trial, and in March 2026, a jury ordered Meta to pay $375 million for endangering children and misleading the public.

Testimony from that trial reportedly showed that Meta's shift toward AI-driven moderation flooded the system with low-quality reports. This made it harder for law enforcement investigations to progress. The verdict should have served as a wake-up call for the entire industry. But the latest investigation suggests that fundamental change is still not happening.

Is This a System Failure or a Calculated Choice?

At some point, describing this as a recurring failure gives Meta too much credit. Every incident follows the same script. Independent researchers or journalists find problematic content. Meta removes it, but only after being publicly confronted. Then the company releases a large removal statistic to argue that it takes the issue seriously. And then the cycle resets.

After nearly a decade of repetition, and with Meta's own data reportedly showing that it can identify duplicate violations without acting on them, the conclusion becomes difficult to avoid. Meta has apparently calculated that the cost of getting caught is lower than the cost of fixing the underlying system.

Several factors make this moment especially alarming. The rapid growth of generative AI tools has made it easier to create realistic abusive imagery at scale. Companies like Meta are now caught between the need to moderate an overwhelming volume of content and the reality that their automated systems are failing. Meanwhile, researchers continue to find sophisticated networks of bad actors who use legitimate advertising tools to distribute harmful content.

There are no simple solutions. Ad platforms need more robust pre-screening processes for advertisers. They need to invest in technology that can detect and remove AI-generated abuse before it is published. And perhaps most importantly, they need to be held accountable with meaningful consequences when they fail. A $375 million verdict should be a warning. Yet the repeated pattern of investigative findings followed by corporate statements and policy tweaks suggests that the industry is still resistant to real change.

For now, users are left with yet another carefully worded statement from Meta. But what is needed is not another statement. It is a system that catches harmful content in the first place. Until that system exists, every promise to take child safety seriously will remain exactly what it has always been: a placeholder waiting for the next investigation to prove it wrong.

Source: Digital Trends News