Meta added an AI system that checks the page an ad links to, not just the ad itself, to catch what it calls “signposting,” ads that look harmless but steer users to CSAM on other sites. Meta can then block the destination and act against whoever ran the ad. The company is also running an AI agent that probes its own safety systems for gaps and is getting better at spotting banned users who come back with new accounts. Meta said it took action on 33.2 million pieces of child exploitation content across Facebook and Instagram from January through June, and that its systems caught more than 97% before anyone reported it. The tools arrive two months after a Wired investigation that found Meta had approved and run ads containing AI-generated child abuse imagery.
