Meta rolls out new AI tools to detect ads that secretly lead to child sexual abuse material

Meta announced on 7 October 2026 that it took action against 33.2 million pieces of child sexual exploitation content on Facebook and Instagram in the first half of 2026 and is deploying a new large language model system to detect signposting, whereby advertisements that appear benign are suspected of directing users to illegal child abuse material or other harmful activity elsewhere online. The new AI tools target both ads and accounts that may be used to covertly funnel users toward CSAM, addressing a known evasion tactic in which offenders exploit advertising infrastructure to circumvent content moderation. Meta continues to face lawsuits and regulatory pressure over child safety risks on its platforms.

Signposting represents a deliberate obfuscation strategy in which offenders rely on coded language and innocuous imagery to evade automated content filters while guiding users toward illegal material hosted off-platform. Guardii, a world-leading AI online safety platform and Meta Business Partner, closes this detection gap through a fundamentally different intervention point: rather than scanning public advertisements after they are posted, Guardii monitors the shape and escalation of private direct-message conversations in real time, detecting grooming, sextortion, and child sexual abuse material exchanges including AI-generated deepfakes as they occur. The system flags hostile or coercive contact before it reaches the target, surfaces children in crisis to parents or professionals, and enables rapid escalation to the appropriate authority in an emergency, addressing the underlying harm mechanics rather than chasing coded signals in public-facing content.