In the space where corporate systems meet the most vulnerable, Meta's advertising infrastructure has repeatedly failed to prevent AI-generated child sexual abuse material from circulating as paid content on Facebook and Instagram — including in India, where the government had already issued a direct removal order. A ten-month investigation by the Tech Transparency Project documents 332 such ads, the majority appearing in a single month, with some placed through Meta's own Chinese advertising partners. The pattern that emerges is not one of isolated failure but of a moderation architecture that
Meta's Ad System Continues Hosting Child Abuse Material Despite Government Orders
Children's images were manipulated using AI to create fake sexual abuse material and distributed as advertising content, causing direct exploitation of real minors.
The system doesn't catch it, but journalists asking for comment gets immediate action.
Mark
So Meta says they've built strong defenses and that most of these ads had fewer than 200 impressions. Does that actually matter if the ads existed at all?
Mimi
It matters for scale, but not for the core problem. The point isn't that millions of people saw these ads. The point is that Meta's moderation system—which is supposed to catch this before anything goes live—failed repeatedly. And when people reported the ads through Meta's own system, the company said they didn't violate policy.
Luke
Right, but we should be precise about what we know. Meta says 43 ads were rejected by their system when TTP reported them. We don't know if those were false negatives—actual violations the system missed—or if TTP's reports were incomplete or unclear. The company claims many had already been removed.
Mark
But TTP is a credible research group. They documented specific ads with specific details.
Mimi
They did. And when TTP contacted Meta directly with the full report, Meta removed all the ads. That's the pattern that keeps repeating: the system doesn't catch it, users can't report it effectively, but journalists or researchers asking for comment gets immediate action.
Luke
Which suggests Meta can remove this content when it chooses to, but the automated systems and the user-facing reporting tools aren't designed to catch it at scale. That's a design question, not just a moderation question.
Mark
What about the Chinese ad resellers placing these ads? That seems like a separate vulnerability.
Mimi
It does. These are companies that channel billions in advertising to Meta. They didn't respond to TTP's questions. We don't know if they knew what they were advertising, if they were deliberately exploiting a gap, or if they were themselves being used by bad actors.
Luke
And that's important to name: we have a theory about what happened, but we don't have confirmation from the resellers themselves. Meta says it prohibits nudify apps, but the ads were running anyway. That could mean the resellers didn't disclose what they were advertising, or it could mean Meta's review of reseller ads is weaker than its review of direct advertisers.
Mark
The Indian government ordered Meta to remove this content two months ago. How is this still happening?
Mimi
That's the question Indian child rights groups are now asking in court. They're arguing that Meta should be reporting directly to Indian law enforcement, not just to a U.S. nonprofit. The government order apparently didn't come with enforcement mechanisms that stopped the ads.
Il Polso
- Hundreds of AI-manipulated images of real children were distributed as paid advertisements on Meta's platforms, with 274 appearing in August alone — a surge that suggests the problem is accelerating, not receding.
- India's government issued a direct removal order after a BBC investigation, yet 78 of the offending ads ran after that directive and after Meta officials had already met with Indian authorities to discuss compliance.
- Meta's own automated reporting system rejected nearly half the abuse reports filed against these ads, telling researchers the content did not violate advertising standards — while the company acted swiftly only when journalists made direct contact.
- Some of the ads were placed not by anonymous bad actors but by Meta's own Chinese advertising reseller partners, raising questions about how deeply the exploitation is embedded in the platform's commercial relationships.
- Indian child rights organizations have now escalated to the Supreme Court, demanding that platforms be legally required to report abuse material to local law enforcement rather than routing all disclosures through a U.S.-based clearinghouse.
- As of the first days of September, new violations were still appearing — 17 documented in a single 48-hour window — signaling that removal efforts remain reactive rather than preventive.
In the space where corporate systems meet the most vulnerable, Meta's advertising infrastructure has repeatedly failed to prevent AI-generated child sexual abuse material from circulating as paid content on Facebook and Instagram — including in India, where the government had already issued a direct removal order. A ten-month investigation by the Tech Transparency Project documents 332 such ads, the majority appearing in a single month, with some placed through Meta's own Chinese advertising partners. The pattern that emerges is not one of isolated failure but of a moderation architecture that responds to journalists faster than it responds to children in danger.
Meta's advertising systems have continued to carry AI-generated child sexual abuse material on Facebook and Instagram even after India's government ordered the company to remove it. A report from the Tech Transparency Project, a Washington-based research group, documents 332 paid ads featuring such imagery over ten months, with 274 appearing in August alone — the majority of them in India, and 78 running after Meta had already met with Indian officials to discuss the problem.
The ads followed a recognizable pattern: real photographs of children, including images of minors from European royal families, were manipulated using artificial intelligence to depict graphic sexual acts. As the footage played, voiceovers promoted AI image-generation apps — many developed by Chinese companies — with language implying users could generate or access abuse material. One ad showing a preteen girl was posted by more than fifty different accounts and ran in India for two weeks.
When researchers attempted to report 84 of the ads through Meta's own system, the company's automated response rejected 43 of them as non-violations. Meta only removed the content after being contacted directly by the Tech Transparency Project — the same pattern observed during the BBC's earlier investigation. The company's public response noted that no moderation system is perfect and that total ad spending on the flagged content was under $5,000. What it did not address was the revelation that several ads had been placed by Meta's own Chinese advertising reseller partners.
The failure has galvanized Indian child protection advocates. Bhuwan Ribhu of Just Rights for Children, a network of more than 250 organizations, has filed a petition in India's Supreme Court arguing that Meta's practice of reporting abuse only to a U.S.-based clearinghouse violates Indian child protection law. Cybercrime officials in Telangana state have pointed to the ease with which perpetrators migrate between platforms as one of the greatest obstacles to prosecution. New ads were still appearing as recently as September 1st and 2nd, with 11 of 17 documented violations that weekend originating in India.
Meta's advertising system has continued to host and distribute artificial intelligence-generated child sexual abuse material across Facebook and Instagram, even after the Indian government explicitly ordered the company to remove such content. A report published this week by the Tech Transparency Project, a Washington-based research group, documents 332 paid advertisements featuring child sexual abuse imagery that ran on Meta's platforms over the past ten months. The majority—274 of them—appeared in August alone.
The findings arrive two months after a BBC investigation revealed that Instagram was actively running paid ads promoting child sexual abuse material in India. That reporting prompted the Indian government to issue a direct order to Meta demanding removal of all such ads and related content. Despite this intervention, the new report shows that 84 of the 332 ads ran in India, and 78 of those appeared after the government's order and after Meta officials met with Indian authorities to discuss the issue. The company's moderation systems, which are supposed to review all ads before they go live, failed to catch the material or allowed it to run anyway.
The ads operated according to a consistent pattern. Nearly all began with a photograph of a real child—in some cases, images of minors from European royal families—which were then manipulated using artificial intelligence to depict the child in graphic sexual acts. As the video played, a voiceover would promote AI image-generation apps, many developed by Chinese companies, with language suggesting they could be used to create or access child sexual abuse material. One ad that circulated in India over a two-week period in August showed a preteen girl animated to perform a sexual act, accompanied by a voiceover claiming "there are no restrictions" and that "all the characters are real people." This single ad was posted by more than fifty different accounts.
When the Tech Transparency Project attempted to report 84 of the Indian ads through Meta's own reporting system, the company's automated response rejected 43 of them, stating they did not violate advertising standards. For 11 others, Meta said the content had already been removed. Only one report remained unanswered. When TTP contacted Meta directly about its findings, the company removed all the ads it had reported. This mirrors what happened during the BBC's original investigation: Meta did not act on abuse reports submitted through its standard channels, but moved quickly once journalists asked for comment.
Meta's statement to the BBC emphasized that "no system is perfect" and that criminals constantly adapt their tactics. The company noted that many of the flagged ads had fewer than 200 impressions and that total spending on them was under $5,000. But the report revealed a more troubling detail: several of the ads featuring child sexual abuse were placed by Meta's own advertising partners in China—resellers who channel billions of dollars in Chinese advertising onto Facebook and Instagram each year. These partners did not respond to requests for comment.
The failure has prompted action from Indian child protection advocates. Bhuwan Ribhu, founder of Just Rights for Children, a network of more than 250 organizations working to prevent violence against children, argues that Meta's practice of reporting abuse only to the U.S.-based National Center for Missing and Exploited Children falls short of Indian law. His organization has filed a petition in India's Supreme Court, citing the BBC's findings and asking the court to mandate that social media platforms report child sexual abuse content directly to Indian law enforcement agencies. "Social media companies are clearly flouting Indian laws by not reporting child sexual abuse content to Indian law enforcement as mandated by our child protection laws," Ribhu said.
Shikha Goel, director of the Cyber Security Bureau in India's Telangana state, told the BBC that one of the largest obstacles to stopping this abuse is the ease with which perpetrators move between platforms, making investigations and prosecutions difficult. She expressed hope that the Indian government would use its ongoing interactions with social media companies to address the problem directly. The distribution of child sexual abuse material is a criminal offense in India, and Meta's own advertising policy explicitly prohibits content that sexually exploits or endangers children. Yet as of early September, new ads continued to appear on the platforms, with 17 additional violations documented on September 1st and 2nd alone, including 11 in India.
Citazioni salienti
Social media companies are clearly flouting Indian laws by not reporting child sexual abuse content to Indian law enforcement agencies as mandated by our child protection laws.— Bhuwan Ribhu, founder of Just Rights for Children
One of the biggest challenges with CSAM is the ease with which users can move from one platform to another, making it harder to nab them.— Shikha Goel, director of the Cyber Security Bureau in Telangana, India
Vuoi la storia completa?
Leggi l'originale su BBC News ↗
Technology