Blog

YouTube’s April 2026 post on expanding likeness detection

Published

YouTube blog — April 2026 update on expanding likeness detection

On 21 April 2026, the YouTube team published an update on the company’s blog about expanding likeness detection to more of the entertainment industry. The piece is aimed at people who work with or appear in professional entertainment — for example through agencies and management — and it describes how YouTube is extending access to a tool that concerns AI-generated or synthetic use of a person’s appearance in content on the platform. Everything below is a paraphrase for readers; for exact language, use the link in the “Sources” section.

What the post covers

In the update, YouTube states that it is broadening the reach of likeness detection in connection with the entertainment business — specifically naming a role for talent agencies, management companies, and the public figures they represent. The post explains that the feature is designed to help surface AI-generated or synthetic content that implicates a participant’s likeness (they give a face-related example on their page) and to support paths to find and request removal under their existing documentation, which they point to from the article. They also compare the general idea to the kind of automated matching that creators and rightsholders already associate with workflows in the Content ID family — that phrasing and scope are on their blog, not reproduced here verbatim.

Partners and who can use the tool

The post names industry partners to illustrate the entertainment rollout (e.g. CAA, UTA, WME, and Untitled Management on that page) and states that celebrities and entertainers may be eligible to use the program even if they do not have a YouTube channel — check the live post and help articles for the precise eligibility and signup process, as those details can be updated.

Partner names, dates, and policy limits may change; always read the current version on YouTube’s or Google’s sites.

Sources (read the originals)