Meta continues to face scrutiny over its handling of child sexual abuse material after a new report found that paid advertisements containing AI-generated abuse imagery continued to run on Facebook and Instagram in India, even after the Indian government ordered the company to remove such content.
The Washington-based Tech Transparency Project identified 332 advertisements containing child sexual abuse material that ran across Meta’s platforms during the past 10 months. Of those, 274 appeared in August alone, according to the report.
India accounted for 84 of the advertisements, nearly a quarter of the total, with 78 of those appearing after the Indian government had issued its removal order and subsequently held a meeting with senior Meta officials.
The findings add to concerns about whether Meta’s systems are capable of detecting and removing abusive material quickly enough, particularly as artificial intelligence makes it increasingly difficult to distinguish between manipulated imagery and authentic photographs.
The report said nearly 80 per cent of the 332 advertisements, or 262, ran in the United States, while 120 appeared in Australia. The Australian advertisements emerged as the country was investigating Meta’s enforcement of its social media restrictions for children under 16.
The latest findings follow a BBC Eye investigation in July that reported Instagram was carrying paid advertisements promoting child sexual abuse material in India. According to that investigation, search terms used in the advertisements directed users towards channels on the Telegram messaging app, where the material could allegedly be purchased for as little as Rs 99, or about 80p.
The allegations prompted action from Indian authorities. The National Commission for Protection of Child Rights issued a notice to Meta demanding an explanation. After reviewing the company’s response, the commission began a formal inquiry and summoned Meta’s managing director and country head to provide further information.
Separately, the National Human Rights Commission instructed Delhi police to conduct a full investigation into the allegations and examine whether Meta and other companies had complied with mandatory reporting obligations under India’s Protection of Children from Sexual Offences Act.
The human rights commission opened proceedings against Meta in July after receiving a complaint.
Meta has rejected the suggestion that it knowingly facilitates or targets such material. Following the BBC investigation, the company said it was “categorically inaccurate” to suggest that it knowingly and deliberately targeted advertisements featuring children at people based on an inappropriate interest in children.
The company said it uses technology to identify accounts displaying potentially suspicious activity involving children and that it had automatically removed more than four million such accounts during the previous year.
But the Tech Transparency Project said its own investigation had identified several images of real children that had been manipulated using artificial intelligence for the advertisements. Among them, it said, was an image of a member of a European royal family.
The group also said more than a dozen advertisements remained live after it had alerted Meta to its findings.
Most of the advertisements identified by the organisation directed users towards AI image or video-generation applications, many of them developed by Chinese companies. Several advertisements were placed through Meta’s own advertising resellers in China, according to the report. Those resellers did not respond to questions from the campaign group.
The findings have raised a broader question about where responsibility lies when abusive material is transformed into commercial advertising and distributed through a major technology platform.
“Content involving child sexual abuse is among the most serious that can be found online,” Katie Paul, the group’s director, said, according to Bloomberg. “But this content isn’t hiding in dark corners of the internet. It’s being used in paid ads on Meta platforms.”
Meta said it does not tolerate “nudify apps or any kind of child exploitation”, whether the material is real or AI-generated. The company argued that the presence of some advertisements did not demonstrate an absence of enforcement, saying that criminals continually change their tactics to evade detection.
According to a Meta spokesperson, many of the advertisements flagged by the researchers had already been removed, most had received fewer than 200 impressions and the total advertising expenditure involved was less than $5,000.
The company said it had built systems designed to block violating advertisements while also conducting ongoing monitoring to identify material that its initial systems might miss.
Yet the Tech Transparency Project said its experience with Meta’s reporting system presented a different picture.
Researchers said they reported the advertisements through Meta’s own system by identifying them as involving someone under 18 and appearing to involve sexual exploitation. According to the report, Meta took at least a week to review the complaints and frequently took no action.
In 57 per cent of the cases reported to Meta, the company allegedly responded that it had reviewed the advertisement and determined that it did not violate its Advertising Standards.
In the remaining 43 per cent, Meta reportedly said the advertisement had already been removed. However, the group said six of those advertisements remained visible in Meta’s Ad Library.
The researchers said the advertisements were taken down only after they contacted Meta directly.
The group subsequently reported all 332 advertisements to the US-based National Center for Missing and Exploited Children, which serves as a centralised reporting system for online child sexual exploitation.
The dispute exposes a significant tension in the rapidly developing AI ecosystem. Meta says its systems are designed to detect abusive activity and that criminals continually adapt their methods to evade enforcement. Researchers, however, argue that the continued appearance of such advertisements — including after direct warnings and government intervention — points to weaknesses in the systems intended to prevent them from reaching users in the first place.
For Indian authorities, the issue is now moving beyond questions about individual advertisements. Formal inquiries by child-protection and human-rights bodies are examining whether technology companies have fulfilled their legal and reporting responsibilities.
For Meta, meanwhile, the challenge is not simply to remove abusive material after it has been identified, but to demonstrate that its advertising infrastructure can prevent such material from being promoted and monetised at all.
The findings from the Tech Transparency Project suggest that, despite Meta’s claims of strengthened defences and extensive enforcement, the gap between detection and removal remains a serious concern. The fact that researchers say some advertisements persisted after they had been reported directly to the company makes the question harder to dismiss as a problem confined to the hidden corners of the internet.
The central issue is therefore no longer whether platforms encounter abusive AI-generated material. It is whether their systems can reliably prevent that material from becoming advertising content in the first place — and whether government intervention is capable of forcing meaningful change when those systems fail.
