Meta was slow to remove hundreds of adverts containing AI-generated child sexual abuse material from Facebook and Instagram, some of which used real photographs of children, including a member of a European royal family and a preteen social media influencer. The findings, published by the Tech Transparency Project (TTP) in a Tuesday investigation alongside separate reporting by Wired, raise serious questions about Meta's ad-vetting systems and come shortly after the company agreed to an $18 billion settlement requiring improvements to child safety on its platforms.

TTP identified 332 ads this year featuring AI-generated CSAM, mostly promoting Chinese-made "nudify" apps that digitally strip or sexualise images of real people, overwhelmingly depicting preteen girls with captions boasting of "no restrictions". This followed an earlier batch of 53 ads flagged in August, after which Meta claimed to have deployed new detection systems, yet TTP found hundreds of further violating ads ran regardless. When TTP flagged the new batch, Meta removed roughly half within hours but initially failed to log many as breaching its child exploitation policy until challenged, while dozens of CSAM ads, including ones using the royal child's image, continued running for days afterwards. Wired separately found hundreds of similar ads sexualising adults without consent, prompting Campaign for Accountability's Michelle Kuppersmith to say Meta's entire ad review system "should be called into question".