Meta adds AI tools to detect child exploitation ads

Meta says it is adding artificial intelligence tools to detect advertisements that appear harmless but may direct users to child sexual exploitation material hosted outside Facebook and Instagram. The company announced the measures on October 7, alongside figures showing it took action on 5.3 million pieces of such content in India and 33.2 million globally on the two platforms from January through June 2026. Meta said more than 98% of the India content and more than 97% of the global content was identified and addressed before users reported it. The figures and detection rates come from the company; the information provided does not include an independent audit or details on how Meta counted content or calculated proactive detection. The new measures focus in part on ads that do not themselves appear to contain illegal material but may direct people to prohibited content elsewhere. Meta said it has introduced a large language model system to identify this kind of suspected signposting and expanded its review of the destinations ads lead to. The company said it can block links to destinations that violate its rules and take action against the accounts responsible.