Meta ads reused AI child abuse material after removal

An investigation by the US nonprofit Tech Transparency Project found at least 300 advertisements containing AI-generated child sexual abuse material on Meta’s platforms in 2026, including ads that used images of real children. DR reports that some of the ads reappeared after being removed.

The investigation, published on 8 September, identified 332 such advertisements in Meta’s ad library. Most of the ads directed users to apps developed by Chinese companies, many of which could generate images or videos using artificial intelligence.

The report found several cases in which images of real children taken from social media had been manipulated into sexual abuse material. It did not specify how many cases involved real children.

One example involved a minor from a European royal family. An image posted on social media by the family was manipulated into a video depicting the minor performing an explicit sexual act. TTP also found an advertisement that turned an image of a 14-year-old influencer into abuse material.

AI-generated child sexual abuse material is illegal in Denmark in the same way as material involving real abuse.

DR’s verification programme, DR Verificerer, reported on Friday that images of children were being used in deepfake sexual videos appearing in advertisements on Meta’s platforms. One of the victims was a Danish minor whom DR was able to identify.

TTP said it found multiple advertisements featuring abuse material involving the same children. The ads appeared repeatedly on Meta’s platforms, even after they had been removed.

“This means that even when Meta had previously identified an advertisement as violating its rules on the sexual exploitation of children, the company did not appear to track the advertisements or images,” TTP director Katie Paul told DR.

According to TTP, Meta had removed 183 of the 332 advertisements by 1 September. After TTP shared its investigation with Meta on 2 September, the company removed the remaining 149 advertisements within a few hours.

However, TTP said it found more advertisements featuring the same video clips one day after the investigation was published on 8 September. Paul said one of the children appeared to be as young as five.

Meta has previously told the BBC and Wired that it “does not tolerate nudify apps or any exploitation of children, whether real or AI-generated”.

The technology company declined to be interviewed by DR but said in an email that it was continuously working to remove such content from its platforms. Meta said it had started legal proceedings against one company for repeatedly advertising nudify apps, issued cease-and-desist demands to more than 45 others, and blocked 92 million sexual advertisements over the previous 10 months.

“The reality is that the criminals behind this activity are constantly changing their tactics to avoid detection, which is why we continue to improve our defences and hold them accountable,” Meta wrote to DR.