Meta Removes 5.3M Child Abuse Posts in India With New AI

Meta just removed 5.3 million pieces of child sexual exploitation content from Facebook and Instagram in India during the first half of 2026, with new AI tools catching nearly all of it before users even reported the material. The breakthrough marks a major win in the fight to protect children online.

The social media giant found and addressed more than 98% of harmful content in India before receiving a single report. Globally, Meta removed 33.2 million such items across both platforms between January and June, with similar proactive detection rates.

The company's new approach tackles a disturbing trend where predators shifted to using advertisements as cover. These ads looked innocent on the surface but secretly directed people to illegal material hosted on other websites.

Meta's upgraded AI system now uses large language models to detect "signposting," which means spotting content that appears harmless but likely leads to illegal material. The technology examines not just what an ad shows but where it actually takes people.

The company also deployed a clever red-teaming AI agent that tests its own defenses for weaknesses, like a friendly hacker finding security gaps before bad actors can exploit them. Additional systems now catch banned users who try to create new accounts.

The Ripple Effect

Meta's technology sharing makes the impact even bigger. The company shares digital fingerprints of identified harmful material with other tech companies through the Tech Coalition's Lantern program, helping protect children across multiple platforms.

Meta's investigative team includes former FBI personnel who work to dismantle entire networks of predatory accounts rather than playing whack-a-mole with individual profiles. The company reports confirmed cases to the National Center for Missing and Exploited Children in the US and now reports directly to India's National Cyber Crime Reporting Portal.

The improvements came after India's government demanded action following a BBC investigation that found paid ads promoting harmful material on Instagram. The Ministry of Electronics and Information Technology summoned Meta executives and made clear such content violates the law without exception.

Meta continues using PhotoDNA and other image matching technology it has deployed since 2011, combining older proven methods with cutting-edge AI systems. This layered approach creates multiple safety nets to catch what individual systems might miss.

The company emphasized that protecting young people remains its most important work, an ongoing effort that never stops evolving as threats change.