YouTube Expands AI Likeness Detection to Combat Deepfakes and Protect Privacy

Generative AI has lowered the barrier to creating deepfakes, allowing the faces of politicians, journalists and celebrities to be misappropriated and threatening election information, personal privacy and public trust. YouTube has responded by developing a likeness-detection system modeled on its Content ID copyright-matching technology, enabling people to identify videos that may use their faces without authorization and request their removal.

YouTube recently expanded the pilot from government officials, political candidates, journalists and Hollywood celebrities to all creators aged 18 and older. The system scans for suspected AI-generated likenesses and alerts the people depicted, but videos are not removed automatically. YouTube will continue to decide whether to remove content through its privacy complaint process, weighing factors including parody, commentary, newsworthiness and the public interest.