Meta acted on 5.3 mn pieces of child abuse content in H1'26; rolls out new AI tools to check ads

Meta disclosed on Wednesday that it took action on a staggering 5.3 million pieces of child sexual exploitation content in India during the first half of 2026. Over 98% of the content on Facebook and Instagram in the country was identified and addressed before being reported by users, according to the company's blog post. Globally, Meta actioned 33.2 million such items between January and June, with a 97% proactive detection rate.

Predators have increasingly turned to advertising channels to hide illicit content, with ads appearing harmless but leading users to illegal material hosted off-platform. In response, Meta has expanded its investigation to include these ads beyond the ones reported to the platform. The company disabled the accounts behind the illicit content and blocked the associated off-platform links.

To combat the growing threat, Meta has deployed new AI tools. A large language model system now detects signposting - content that appears benign but is strongly suspected of directing users to illegal material. The company has also improved its assessment of where an ad leads, considering not only the ad's content but also its destination. Additionally, Meta has introduced AI-driven sweeps to surface material that previous systems might have missed.

An AI red-teaming agent has been created to probe Meta's own defenses for weaknesses, while the company has strengthened detection of banned users who create new accounts. Safety remains a top priority for Meta, with the company stating that it never stops working to keep young people safe on its platforms.