Under regulatory scrutiny in India, social media giant Meta claims to have taken action against 5.3 Mn pieces of child sexual abuse material (CSAM) in the first half (H1) of 2026. 

The social media giant also noted that more than 98% of this content was identified and addressed proactively before being reported. This comes as Meta has come under fire over its lax policies around taking down child sexual abuse content. 

Globally, the social media company said that it took down 33.2 Mn pieces of such content during the same timeframe, with a proactive detection rate exceeding 97%. 

Meanwhile, Meta also said that it has rolled out new AI-powered systems and upgraded its ad review frameworks across Facebook and Instagram in India to detect and take down CSAM. 

The social media juggernaut also said that it has introduced LLM-driven systems to identify “signposting” within ads to counter predators who advertise illicit material in ways that can evade automated detection. 

For context, signposting acts as a roadmap to guide and enable users to anticipate what happens next. Meta’s new “signpost” safeguards evaluate where an advertisement leads rather than solely analysing its visible content. 

Furthermore, the social media company has also deployed AI-driven sweeps to uncover missed CSAM material. It has also integrated “red-teaming” AI agents to probe its own defenses and strengthen detection mechanisms against banned users creating new accounts. 

It is pertinent to note that red teaming AI agents intentionally probe autonomous AI systems or chatbots to find security flaws and safety gaps before malicious actors do.

Source link