What happened
A joint investigation by Futurism and Gizmodo revealed that Facebook was hosting a network of accounts that posted AI‑generated videos depicting older male coaches inappropriately touching young female athletes. The videos showed coaches running their hands over backs, inner thighs, and in some cases, genital areas, as well as kissing the athletes. After Futurism used a regular Facebook account to report ten of these pages, the platform responded a week later that eight of the accounts did not violate its Community Standards and would remain online, while two pages disappeared without explanation. When Futurism followed up with Meta’s media relations team, the eight previously cleared accounts were removed, and the earlier notifications were updated to acknowledge rule violations. Meta confirmed the removals in a statement to Futurism, emphasizing that abusive or exploitative content—real or AI‑generated—has no place on its services.
Details of the content and moderation response
The AI‑generated clips were described as “nearly photorealistic” and featured gymnasts, ballet dancers, and yoga practitioners wearing leotards. The athletes appeared to be teenagers, some shown with braces, and reacted with smiles or giggles despite the abusive behavior. Some videos framed the touching as coaching instruction, while others were overtly sexual, including scenes of coaches touching genitals over clothing or sitting on a gymnast’s lap.
Futurism reported that individual videos attracted over 100,000 views each. One account posted multiple variations of the same scenario and used the comments to promote a “members only” WhatsApp channel for additional content. Another account’s header image displayed a coach’s hand on a gymnast’s crotch.
When the reports were filed, Facebook sent identical notifications to each of the eight accounts, stating that the pages did not breach Community Standards. The messages quoted Meta’s policy language about using a mix of technology and human reviewers. After the media inquiry, the same accounts were taken down, and the notifications were revised to indicate that the pages did violate the rules.
Meta’s official response highlighted its broader moderation performance, noting that in the previous quarter the company proactively removed 98 % of child‑endangerment posts before they were reported. The company also referenced its Community Guidelines, which forbid “impending sexual activity” and “adult sexual exploitation” in both real and digitally created imagery, including AI‑generated material.
Background on Meta’s policies and AI moderation efforts
Meta’s Community Standards outline prohibitions for adult nudity, sexual activity, and child sexual exploitation. The guidelines specifically mention that digitally created or AI‑generated images that are sexually suggestive are not allowed. In its latest Community Standards Enforcement Report, Meta disclosed that early tests of large language models (LLMs) reduced moderation mistakes by 13 % compared with human reviewers and identified 10 % more actual violations. The company is therefore experimenting with AI to assist its moderation teams, although the effectiveness of these tools remains under evaluation.
Implications for content moderation and AI‑generated abuse
The incident underscores the difficulty social platforms face in policing AI‑generated content that mimics real‑world abuse. While Meta claims high removal rates for child‑endangerment material, the initial clearance of the coach videos suggests gaps in the detection pipeline, especially when the content is generated by AI rather than captured in the real world.
Similar challenges have emerged on other platforms. Earlier in the year, Elon Musk’s X faced scrutiny after its AI chatbot Grok was used to create sexualized images of people without consent, including minors. These cases raise broader questions about whether existing moderation frameworks can keep pace with the rapid proliferation of synthetic media.
For users and advocates, the episode highlights the importance of external scrutiny. Futurism’s reporting prompted Meta to act, demonstrating that journalistic inquiry can serve as a check on automated moderation systems that may otherwise miss harmful material.