YouTube just expanded its AI likeness detection tool to all adults over 18, making it one of the first major video platforms to offer automated deepfake protection to ordinary users.
The feature works through a selfie-style face scan. You submit your image, YouTube's AI scans the platform for videos featuring people who look like you, and you get an alert when a potential match is found. From there, you can request removal of videos that use your likeness without permission.
The expansion matters because the tools to create convincing AI-generated video using someone else's face are cheap and widely available now. Previously, a non-celebrity who discovered their face in an AI-generated video had almost no practical recourse - manually searching YouTube for yourself isn't realistic. Automated scanning changes that.
The tool's real-world usefulness depends on two things YouTube hasn't addressed publicly: how accurate the face matching is, and how quickly removal requests actually get processed. Face matching at YouTube's scale - 500 hours of new video uploaded every minute - is technically hard, and false positives could overwhelm the system's usefulness. It's also opt-in, which means the people most at risk from deepfakes (those without large followings who won't hear about this launch) are the least likely to find the setting.
Still, the direction is correct. Platforms hosting AI-generated content bear some responsibility for helping subjects identify and report misuse. A face-matching scan is a more practical tool than asking users to police a platform this size on their own.