Meta Says It Acted on 33.2 Million Child Exploitation Items in First Half of 2026, Adds AI Screening for Ads That Point to Abuse Material

Since India's IT ministry ordered Meta in July to disable Instagram ads tied to child sexual abuse material, and since this week's coverage of a U.S. appeals court ruling on AI-generated abuse images, the focus has shifted to what the platforms are doing about the problem. On Wednesday, Meta gave its answer in a blog post.

The numbers

Meta said it took action against 33.2 million pieces of child sexual exploitation content on Facebook and Instagram between January and June 2026. More than 97% of it was found by Meta's systems before any user reported it.

India accounted for 5.3 million of those items, roughly 16% of the global total. Meta said more than 98% of the Indian content was caught proactively.

These are Meta's figures. "Actioned" counts pieces of content, not people or arrests, and the company did not say how many items were ads.

The ad problem

Meta said predators have moved toward advertising. An ad can look clean on its own while sending people to illegal material hosted on outside websites. Meta calls the tactic "signposting."

The company said that once it identified the scheme, it widened its investigation beyond the ads that had been reported to it, disabled the accounts behind them and blocked the off-platform links.

The new tools are:

• A large language model system built to flag signposting, meaning ad content that looks benign but is strongly suspected of directing people to illegal material.
• Analysis of an ad's destination, so Meta can block violating websites and act on the accounts behind them.
• Additional AI sweeps meant to catch exploitation content that earlier systems missed.
• A "red-teaming AI agent" that probes Meta's own defenses for weaknesses.
• Stronger detection of banned users who come back under new accounts.

Meta also said it still relies on PhotoDNA and other image and video matching, which it has used across its apps since 2011. It shares hashes of identified material with other companies through the Tech Coalition's Lantern program and reports confirmed material to the National Center for Missing & Exploited Children. Its investigators, who include former FBI personnel, target networks rather than single accounts, the company said.

"Keeping young people safe on our apps is among the most important work we do, and it never stops," Meta said.

Pressure from New Delhi

The timing follows government pressure. In July, India's Ministry of Electronics and Information Technology issued Meta a notice over such ads on Instagram, ordered them disabled and summoned the company's global executives to India.

Meta announced in September that it would report child safety cases directly to the National Cyber Crime Reporting Portal run by the Indian Cybercrime Coordination Centre (I4C). It said that builds on its existing reporting of CSAM to law enforcement through the portal, with help from NCMEC.

No source says whether the ministry considers the new measures sufficient.

Legal exposure

TechCrunch reported that in August Meta agreed to pay up to $18 billion to settle a child safety lawsuit involving 29 U.S. states. Meta has also added parental controls this year, including controls for Meta AI, preteen accounts on WhatsApp and alerts when children search Instagram for self-harm content.

What the numbers leave open

A 97% proactive detection rate sounds strong. It is also a share of what Meta found, and no one outside the company has verified how much it missed. The tactic Meta describes, harmless-looking ads that route users elsewhere, is built to avoid detection by design.

That is why the destination check matters. Scanning only the ad creative was never going to catch it. Whether the new system catches sites that show reviewers one thing and users another is not something Meta's announcement addresses.

There is also a separate legal gap in the U.S. A federal appeals court ruled recently that AI-generated images depicting no real child are protected speech under the First Amendment, and it urged Congress and state legislatures to update the law. An opinion writer in The Hill argues that synthetic material still harms children. The piece says it can fuel escalating fixation, draw on photos of real children and be used in deepfakes of identifiable victims. The piece cites a Tennessee case in which a man allegedly used Grok to undress teenage girls and spread the images on Discord and Telegram. Meta's tools are built to catch abuse material and the ads that lead to it on its own platforms. They do not change what the appeals court said prosecutors can charge.

Meta said it will keep adding detection signals as it learns more about how these networks operate. No audit date, outside review or reporting schedule for the new tools has been announced.