Meta Faces Scrutiny Over AI-Generated Violent Child-Abuse Videos

Futurist investigators have documented a large network of Facebook accounts that repeatedly upload AI-generated videos showing extreme violence against children. The material includes scenes of infants and toddlers being slapped, kicked, burned, chained, locked in freezers, and subjected to other brutal acts. Some videos feature children of color being beaten by white adults in a colonial-style setting, while others depict nonsensical escalations of violence that shift from child abuse to adults punching each other. The videos are not photorealistic but are clearly produced with advanced generative-AI tools, and they attract thousands of reactions from users who often mistake the content for real footage.

Futurism

The company declined to comment further, directing inquiries to a public page describing its automated child-abuse detection efforts, which does not disclose removal numbers for AI-generated content.

Futurism

The investigation flagged eight accounts that shared the abusive videos. Meta ultimately removed only two of them, and one of those removals required an initial rejection followed by an override. The remaining six accounts stayed online for more than a week, despite containing multiple videos of toddlers being struck with frying pans, girls beaten with baseball bats, and babies kicked. The platform’s recommendation engine continuously surfaces similar material, creating a “bottomless feed” of disturbing content.

Futurism

Meta has invested heavily in AI research, offering multi-billion-dollar packages to attract top talent and claiming a strategic advantage in building “superintelligence.” Independent testing shows that Anthropic’s Claude model can reliably identify child-violence imagery, and internal Meta systems appear capable of detecting the same videos. Nonetheless, the company has not consistently applied these tools to remove the content proactively.

Futurism

The persistence of AI-generated child-abuse videos raises serious concerns about platform safety, the effectiveness of existing moderation policies, and the ethical use of generative AI. It also highlights a pattern in which Meta’s stated commitment to combating harmful content does not translate into decisive action, leaving thousands of disturbing videos accessible to users.

Futurism