Meta announced on Wednesday that it is implementing new AI tools to identify ads and accounts that may appear normal but are used to secretly direct users to illegal child abuse material off its platforms. The company stated that it took action against 33.2 million pieces of child sexual exploitation content on Facebook and Instagram in the first half of 2026, with over 97% detected by its systems before user reports. The new measures include a large language model (LLM) system to detect "signposting" ads, additional AI-driven scans for missed content, a "red-teaming AI agent" to test safety measures, and improved systems for identifying returning offenders. This comes as Meta faces ongoing pressure and lawsuits regarding child safety on its platforms, including an agreement in August to pay up to $18 billion to settle a child safety lawsuit with 29 U.S. states.