Meta has introduced AI-powered tools to detect and remove ads on its platforms that appear harmless but direct users to child sexual abuse material (CSAM) hosted elsewhere online. One tool is a large language model designed to identify “signposting”—benign-looking ad copy suspected of covertly leading to illegal content.
Meta said the measures respond to predators shifting tactics to exploit advertising systems. The company has also improved its analysis of ad link destinations and is conducting additional AI-driven sweeps to find material its previous systems missed.
The announcement comes amid increased scrutiny and government pressure, particularly in India, over the presence of such material on Meta’s platforms.