To police the spread of misleading AI-generated content, YouTube has announced the expansion of its automated likeness detection initiative from a small group of journalists and civic leaders to
entertainment industry accounts, including talent agencies, management companies, and the celebrities they represent.
“Likeness detection works similarly to Content ID: it looks for
AI-generated content with a participant’s likeness, like a deepfake of their face, and gives them the power to find it and request removal,” the company’s announcement explains.
Overall, the detection technology was designed to safeguard creators’
identities while ensuring that their audience is not misled, especially with regard to the brands, products and services creators are purposefully endorsing.
advertisement
advertisement
Moving forward with its detection
of predatory AI-generated content targeting those in the entertainment industry, YouTube has garnered support from major talent agencies and management companies such as CAA, UTA, WME and Untitled
Management.
One of the most egregious examples of a celebrity deepfake endorsement on YouTube took place last year, involving an AI-generated version of Tom Hanks featured in a widespread ad
promoting a so-called cure to reverse type 2 diabetes in less than three weeks.
Hanks issued a statement to the public on Instagram warning them about this “17-second grape trick”
and other similar ads, writing that they had “been created without my consent, fraudulently and through AI.”
Shortly after MediaPost discovered that the Hanks ad continued to run on YouTube for several more months, the
company publicly backed the “No Fakes Act” as well as the “Take It Down Act,” which both aim to tackle the increased spread of unauthorized digital replicas.
During that
time, YouTube expanded its deepfake prevention program, bringing likeness
detection to all 4 million eligible creators in the YouTube Partner Program, enabling them to ask that AI-generated content using their likeness be removed from the platform.
Last month,
YouTube brought its detection tech to a pilot program of leaders at the center of civic discussion, debate, reportage and policy. At the time, YouTube expressed plans to “significantly expand
access” to this tool over the coming months, which now covers celebrities — a highly targeted group for likeness infringement.
Notably, the company has stated that “detection does
not guarantee removal,” but that YouTube will continue to preserve parody and satire, “even when used to critique world leaders or influential figures,” and will “carefully
evaluate these exceptions” when receiving requests for removal.


















