CSAM safeguards: Meta details AI detection, enforcement actions, and ad review changes

Meta on Tuesday described steps it said it took to curb child sexual abuse material across Facebook, Instagram and WhatsApp. The company highlighted AI-led detection, account removals and ad checks. The statement came days after the government issued a notice over Child Sexual Exploitative and Abuse Material in paid Instagram advertisements.

Representative image

In a Tuesday blog post, Meta called child exploitation a horrific crime and said it acted daily on and off its platforms. Meta wrote: "This work is ongoing. Our teams are constantly improving our defences -- developing new technology, blocking violating links, and sharing intelligence across the industry -- but we know there is more to do. We will continue investing in every resource needed to keep young people safe, strengthen our ad review processes, and work with law enforcement to hold criminals accountable,\" Meta said.

Meta child exploitation response and India government notice

Last week, the government issued a stern notice to Meta over CSEAM in paid Instagram ads. The Ministry of Electronics and Information Technology ordered Instagram to disable ads and content. It also asked for a detailed explanation within seven days. The move followed IT Minister Ashwini Vaishnaw asking officials to summon Meta.

Government sources said Meta’s formal response to Saturday’s notice was awaited. They said the government focus would be on corrective steps and actions by the company. The regulatory attention followed a BBC report alleging Meta’s recommendations promoted videos containing abuse material. The report also alleged such ads appeared despite Meta’s ad rules.

Meta AI detection and child exploitation enforcement figures

Meta said it was wrong to claim it targeted ads with children to people with improper interests. Meta said: \"Quite the opposite; we use technology to identify accounts that have shown potentially suspicious activity related to children, and we automatically removed over four million of these accounts last year,\" Meta said. Meta said newer AI systems covered languages used by 98 per cent online.

Meta said it removed over four million suspicious accounts and 36 million pieces of child exploitation content globally last year. In India, Meta said AI tools removed 1,60,000 accounts in six months. Meta said it used signals to detect suspicious off-platform links. It said this activity was linked to exploitative behaviour.

Meta said it knew of reports about Instagram ads in India that broke its child exploitation rules. Meta stated: \"Were aware of recent news reports about Instagram ads in India that violated our policies against child exploitation. And we want to be clear: we take these concerns seriously, we never want this content on our platforms, and were committed to improving our efforts to combat it,\" it noted.

Meta ad review process and child exploitation policy enforcement

Meta said it used automated tools and human reviewers to find and remove ads that broke policy. It said no system could catch every violation. Meta said ads were screened before running and stayed under ongoing review. It also said users could report suspected violations through platform tools.

Meta said it tracked advertiser behaviour as well as individual advertisements. It said it could reject ads and restrict business accounts, ad accounts, pages and user accounts. Meta said it was improving ad review and enforcement to stop bad actors. Meta said: \"Were committed to keeping bad actors off our platforms and are constantly evolving our systems to stay ahead of them. Protecting people who use our platforms remains at the centre of how we build and enforce our advertising standards,\" it said.

Meta said it followed a zero-tolerance approach on child nudity, abuse and exploitation. It said its rules banned sharing or seeking exploitation imagery and sexualisation of minors. Meta said: \"As noted in our Ad Standards, all ads must comply with our Community Standards on Child Sexual Exploitation, Abuse, and Nudity. Ads must not contain content that sexually exploits or endangers children,\" it said.

Meta said its systems had already found and disabled some violating ads and accounts. Meta said: \"Our subsequent investigation led to additional action, including removing further ads, disabling accounts, and blocking URLs linked to policy-violating content,\" it pointed out. Meta said it used AI tools to detect suspicious links shared off-platform.

Meta said it reported apparent cases to law enforcement through the National Center for Missing and Exploited Children NCMEC. Meta said it published global and India-specific transparency reports. It said it had appointed a Chief Compliance Officer, Grievance Officer, and Nodal Contact Person under India’s IT Rules.

Meta said it worked with law enforcement and other firms beyond its own apps. It cited its role in industry efforts, including its founding membership of tech coalitions. Meta highlighted the Lantern programme for cross-platform intelligence sharing. It also said it worked to block links to third-party sites hosting abusive content.

Separately, reports alleged Instagram showed paid ads using terms like rape video and child video. The ads allegedly sent users to Telegram channels where such content was offered for sale. Meta said it aimed to keep such content off its platforms. The company said it would keep investing in technology and resources.

With inputs from PTI