YouTube has begun rolling out its likeness detection tool to help creators identify and report AI-generated videos using their faces without permission. The beta feature, now expanding beyond initial testers, requires identity verification but offers no removal guarantees. This move addresses growing concerns over AI content proliferating on the platform.
Google, which powers YouTube, announced earlier this year tools to flag face-stealing AI content. As of October 21, 2025, the company has started notifying the first batch of eligible creators about access to the likeness detection system, similar to its copyright detection mechanism.
The feature is currently in beta and limited testing, appearing in the 'Content detection' menu of YouTube Studio for select users. Setup assumes a single host per channel and requires verification via a government ID photo and a face video. Once enrolled, the algorithm scans other channels for videos matching the user's face, though it may produce false positives, such as fair-use clips.
Creators spotting potential AI fakes can submit reports with details. If the content copies from their channel without fair use, a separate copyright removal request is recommended. However, YouTube does not guarantee removal. Reviewers consider factors like parody labeled as AI or unrealistic styles, which may stay online, versus realistic videos showing endorsements or illegal activities, which are likely to be removed.
This rollout comes amid rising AI video capabilities. Google recently unveiled its Veo 3.1 model, supporting portrait and landscape formats, with plans to integrate it into YouTube. Rival OpenAI's Sora models have popularized AI videos, often featuring public figures in fabricated scenarios, potentially increasing likeness complaints akin to DMCA takedowns.