Meta Expands Takedown of AI-Generated Child Sexual Abuse Material

Meta has announced enhanced detection and removal policies for AI-generated child sexual abuse material across Facebook, Instagram and Threads. The company will now proactively detect and remove synthetic CSAM created using generative AI tools, even when no real child is depicted, and will disable accounts that repeatedly share such content. Meta is deploying machine learning classifiers trained to identify AI-generated imagery and has added new hashing techniques to prevent re-upload. The policy change follows pressure from child safety advocates and regulators who warned that generative AI is being used to create exploitative images at scale, with the National Center for Missing and Exploited Children reporting a sharp increase in AI-generated CSAM reports in 2026.

Detection at the point of upload addresses distribution but not the private-channel solicitation and coercion that precedes it. Guardii's anti-CSAM module, built to recognise both photographic and AI-generated material, operates in direct messages before content is shared, intercepting requests for images, threats used to extort them, and the grooming conversations in which offenders desensitise targets to sexual content. As a Meta Business Partner, Guardii extends the platform's public-facing controls into the private layer where most exploitation occurs, flagging the conversation that leads to image-based abuse rather than waiting for the image itself to surface, and enabling intervention while the child is still in contact with the offender.