The system operates automatically across uploads, identifying unique facial patterns even when manipulated and routing matches to authorized rights holders for review, without requiring the person to have a YouTube channel.

Actions available for detected content include removing the video for privacy policy violations or submitting a copyright removal request, with allowances made for parody and satire rather than automatic removals.

YouTube is expanding its AI likeness protection to Hollywood, enabling actors and their representatives to flag and flaggable AI-generated content that mimics their appearance.

YouTube plans a scalable, automated, platform-wide monitoring system aimed at proactive protection, though current removal counts remain very small.

Future iterations will add audio analysis to detect AI-cloned voices, expanding protection beyond visuals.

Overall, the initiative signals a shift toward rights management in AI use on platforms, emphasizing consent, contracts, and privacy safeguards.

The move is framed as a landmark step toward a ‘right of publicity’ in the AI era, balancing actor protections with fan content and free expression.

Hollywood agencies like CAA, UTA, WME, and Untitled Management are invited to enroll clients’ faces, reflecting a broader industry adoption to monitor and protect likenesses across the platform.

Industry perspectives view the tool as beneficial for visibility and protection, though concerns remain about preferential access for elites and potential impact on fan-created content.

The expansion builds on a multi-year collaboration that began with CAA and has evolved to include a broader set of high-risk users, including celebrities and public figures.

Enrollment requires identity verification and data used to verify likeness will not be used to train Google’s generative models; participants can opt out and data is deleted upon withdrawal.

Details on eligibility, access, and safeguards are provided by YouTube, with the tool positioned as a proactive measure against misrepresentation rather than an immediate monetization pathway.