YouTube Opens AI Likeness Detection to All Adult Creators

Any creator aged 18 or older can now enroll in YouTube's likeness detection tool, a system that scans newly uploaded videos for AI copies of an enrolled creator's face and flags them for review. The reach is new. The tool launched for Partner Program members in October 2025, after a pilot phase, and opened to all adult creators in May 2026. At the Made on YouTube event in New York on September 23, 2026, the company said it will start folding speaking voice detection into the same system later this year, and that enrollment and match alerts are coming to the mobile app.

How the detection works

YouTube scans new uploads against the face templates of creators who have signed up. When the system finds a likely match, the video shows up in the creator's YouTube Studio, under the Likeness tab in the Content Detection menu. The creator reviews the clip and chooses what happens next: a removal request under YouTube's Privacy Guidelines, a copyright claim, or archiving the match and leaving the video alone. Matches are judged one by one. YouTube weighs factors such as whether the video looks realistic and whether the person can be clearly identified, and it keeps room for parody and satire, so a flag does not guarantee a takedown.

Enrollment asks for a photo ID and a selfie

Signing up starts in YouTube Studio on a computer. The creator opens the Likeness tab, agrees to data processing, and scans a QR code with a phone to open a verification page. YouTube asks for a government issued photo ID and a short selfie video. YouTube's help pages say the company stores the selfie video, the creator's legal name and the likeness template for up to three years after the creator's last sign in. A separate, optional checkbox lets YouTube use face and voice templates to improve its detection models. Creators can opt out at any time, and YouTube says scanning stops 24 hours after someone opts out.

From a Partner Program perk to everyone

YouTube first previewed the technology in 2024 and worked with Creative Artists Agency on early testing with celebrities and creators. Eligibility then widened in steps: journalists, government officials and political candidates in March 2026, celebrities in April 2026, and every eligible adult creator in May 2026. Spokesperson Jack Malon told The Verge that anyone can use the tool, and said first time uploaders and decade long creators would get the same level of protection.

What it still cannot do

For now, detection runs on faces alone. A cloned voice by itself will not trigger a match, and a creator who finds an AI copy of their voice has to report it by hand through the privacy complaint process. YouTube calls the tool experimental, and it is not offered in every country. The voice integration announced in September is meant first to sharpen face matches. YouTube has not said that a voice only fake, with no face on screen, will be caught.

The privacy trade off has been debated since the first launch. MediaNama founder and editor Nikhil Pahwa raised concerns early on about creators having to hand over government IDs and facial data in exchange for protection. In India, courts have repeatedly ordered platforms, YouTube among them, to act against voice cloning, restraining unauthorized use of singer Arijit Singh's voice and likeness in August 2024 and of singer Asha Bhosle's voice in October 2025.