YouTube Expands AI Deepfake Tool to Public Figures

YouTube announced on Tuesday the expansion of its AI-powered likeness detection tool to a pilot program targeting government officials, political candidates, and journalists, aiming to combat unauthorized deepfake videos that impersonate these public figures.

The feature, first introduced last year for creators in the YouTube Partner Program, scans uploaded videos for AI-generated content mimicking a person's appearance. Participants in the new pilot can review flagged videos through an online dashboard and request their removal if the content violates YouTube's privacy guidelines. The system operates similarly to the platform's Content ID technology, which identifies copyrighted material, but focuses on visual patterns indicating synthetic faces rather than audio or clips.

To enroll, eligible individuals must verify their identity by submitting a short video selfie and government-issued identification. YouTube stated that this data will be used solely for verification and to enable the detection feature, and it will not be applied to train Google's generative AI models. Once verified, the tool automatically monitors uploads and alerts users to potential matches, allowing them...