Have you recently found an AI-generated version of yourself in a YouTube video? Don't worry, you can now ask YouTube to take it down.
This comes as YouTube has quietly introduced a new policy addressing concerns over AI-generated imitations of people and allowing affected individuals to request their removal through YouTube's privacy request process. This is the latest in ongoing efforts by YouTube and other social media platforms to address synthetic media and AI content shared by their users.
The rise of AI-generated content poses a major challenge for platforms like YouTube. Synthetic media such as deepfakes can create highly realistic but misleading representations of people, raising privacy concerns and potential misuse. In March, YouTube introduced a tool in Creator Studio for creators to disclose if they've used synthetic media to create their content. The platform is also piloting a crowdsourced notes feature that indicates whether videos with AI content are misleading or simply parody.
“If someone has used AI to modify or create synthetic content that looks or sounds like you, you can request it be removed,” the new policy states. “To be eligible for removal, content must realistically modify or synthesize your likeness.”
The updated policy allows individuals to file complaints about AI-generated content they believe violates their privacy; previously, YouTube would simply label such content as AI-created or misleading.
Privacy issues
Currently, a person portrayed by an AI tool or their lawyers can, at least in most cases, file a privacy complaint to have the person removed. The content creator has two days to remove the likeness or the entire video after receiving the complaint. If they do nothing, YouTube will review the complaint to determine whether it has merit.
However, there is no guarantee that it will be removed: the decision to remove will depend on factors such as whether the video acknowledges its AI origins, whether specific people are identified, whether it was made for parody or satire purposes, or if the public figure is committing a criminal act or endorsing someone or something.
It's worth noting that privacy complaints are separate from Community Guideline violation strikes. A privacy complaint doesn't automatically mean a strike against the user, but repeated privacy violations can get the user banned.