YouTube announced on Tuesday that it is expanding its advanced “likeness detection” technology to the entertainment industry, providing a new layer of protection against AI-generated deepfakes for celebrities and public figures.
How the AI Protection Works
The technology functions similarly to YouTube’s established Content ID system. While Content ID tracks copyright-protected material to allow rights owners to claim revenue or request removals, the likeness detection tool specifically targets simulated faces. This initiative aims to safeguard creators and public figures from having their identities exploited in unauthorized content, a recurring issue for celebrities frequently targeted by scam advertisements.
Evolution and Industry Adoption
Initially launched as a pilot program for a select group of creators last year, the technology expanded this spring to include government officials, politicians, and journalists. The company announced that the tool is now available to talent agencies, management firms, and the celebrities they represent. Major industry players, including CAA, UTA, WME, and Untitled Management, have provided feedback to refine the tool.
Control Over AI-Generated Content
Entertainers do not need to maintain their own YouTube channels to utilize this protection. The system scans for AI-generated content that matches an enrolled individual’s face. Once a match is identified, the rights holder can request removal based on privacy policy violations, submit a copyright claim, or choose to take no action. YouTube maintains that it will not remove all AI content, as its policies continue to permit parody and satire.
Future Developments and Policy Advocacy
YouTube plans to extend the detection capabilities to include audio in the future. Beyond its internal tools, the company is actively advocating for broader protection at the federal level, specifically supporting the NO FAKES Act in Washington, D.C. This legislation aims to regulate the use of AI in creating unauthorized digital replicas of voices and likenesses. While the company has not disclosed the exact number of deepfakes removed via the tool, it noted in March that the volume of such removals remains relatively low.
About the Author
