Meta deploys AI against ads leading to child exploitation material — TechCrunch

As TechCrunch reports, Meta has begun rolling out new artificial intelligence tools to detect ads and accounts that may covertly direct users to illegal child sexual exploitation material outside the company’s platforms.

Meta said it took action on 33.2 million pieces of child sexual exploitation material on Facebook and Instagram in the first half of 2026. According to the company, its systems detected more than 97% of such content before user reports. In India, Meta took action on 5.3 million such pieces of material, more than 98% of which were detected by its systems before user complaints.

Checking advertising links

Meta’s new large language model is intended to detect so-called covert redirection through advertising. This refers to ads that may appear ordinary but, according to the company’s suspicions, direct users to external resources containing illegal or other harmful content.

Meta noted that malicious actors are changing their methods to avoid detection. The ads themselves may not contain illegal material, but may serve as a path to websites where it is hosted.

More current news is available on the UA.News Telegram channel Telegram.

The company will check not only the content of ads, but also where they direct users. According to Meta, this will make it possible to block websites and other resources that violate its rules, as well as take action against accounts associated with them.

Additional safety tools

Meta is also introducing additional AI scanning to search for child exploitation material that previous systems may have missed. The company plans to add new signals as it studies how such networks operate.

Another tool is an AI agent for testing Meta’s own safety mechanisms. It will look for weak points that malicious actors could exploit to bypass protections. The company is also improving systems for detecting people who return to its platforms through new accounts after their accounts have been removed.