The group says it took action against 33.2 million pieces of child sexual exploitation material in the first half of the year and has introduced an AI agent tasked with bypassing its own safeguards.
Meta reported taking action against 33.2 million child sexual exploitation content on Facebook and Instagram in the first half of 2026, of which more than 97% were identified by its systems before any user reports. In India, the figure is 5.3 million content, detected at over 98% proactively.
Watch where advertising leads
The group is putting into operation a system based on a large language model to uncover "signposting": banal-looking ads that actually serve as a gateway to sites hosting child-related content outside Meta's platforms. Since the advertisement itself contains nothing illegal, Meta now analyses the destination to which it is referring, to block these sites and sanction the accounts at the origin of the campaigns. Further analysis by IA must also catch up with content missed by old systems.
Meta adds an IA "red teaming" agent to look for gaps in its own protections before malicious actors exploit them, and improves the detection of people returning to a new account after a ban. These announcements come as the group continues to be under pressure on the protection of minors, after agreeing in August to pay up to $18 billion for a lawsuit brought by 29 American states.
Read the full article on the original source
Go to the source →