Meta initially said there was no violation, then deleted the disturbing AI videos after reporters pressed

A Futurism investigation found that multiple Facebook accounts sharing AI-generated videos remained undeleted after user reports; the videos depicted coaches inappropriately touching young women and girls during gymnastics, ballet and yoga training. Meta initially said 8 of the pages did not violate community standards, and only after Futurism contacted its media team did it remove the accounts and update some of the notices. Futurism reported 10 pages using an ordinary account, and a week later 8 were judged not to violate the rules, while the other 2 had disappeared for unknown reasons. Meta confirmed to Futurism that it had removed the content, saying content depicting abuse or exploitation is not allowed on its platforms whether real or AI-generated, and that in the previous quarter it proactively detected and removed 98% of child-harm posts. Meta is also testing LLM-assisted content moderation, with preliminary tests saying it made 13% fewer errors than humans in judging violating content and found 10% more violating content. The article also mentioned that X's Grok had previously been used to generate non-consensual sexualized images and prompted a government investigation.