YouTube Expands Its Anti-Deepfake Technology to Movie Stars

⚡

Key Takeaways

1YouTube has expanded its deepfake detection technology to include entertainment celebrities.

2This technology works like Content ID, but for faces, protecting public figures from impersonation.

3Talent agencies like CAA and UTA support this initiative, providing feedback on the tool.

💡Why it matters — Protecting celebrities from deepfakes strengthens the fight against digital impersonation, a crucial issue in the AI era.

YouTube Strengthens Its Fight Against Deepfakes by Targeting Celebrities

YouTube recently announced the expansion of its similarity detection technology to include personalities from the entertainment industry. This advancement aims to identify content generated by artificial intelligence, such as deepfakes, that mimic the faces of celebrities. The announcement was made by the company this Tuesday, highlighting its commitment to protecting the image of public figures.

The technology in question is inspired by the existing Content ID system on YouTube. This system is designed to detect copyrighted content in videos uploaded by users. It allows rights holders to request the removal of videos or share the revenue generated. The new feature applies this principle to simulated faces, thus providing additional protection to creators and other public figures against the unauthorized use of their image, a recurring issue particularly in the context of fraudulent advertisements.

Extended Protection to New Sectors

Initially, this similarity detection technology was tested with a small group of YouTube creators as part of a pilot program launched last year. In the spring, it was expanded to include politicians, officials, and journalists. Now, YouTube has decided to extend this protection to those in the entertainment industry, including talent agencies, management companies, and the celebrities they represent. Influential agencies such as CAA, UTA, WME, and Untitled Management have supported this initiative by providing valuable feedback on the tool.

An Accessible Tool Without a YouTube Channel

One of the unique features of this tool is that it does not require artists to have their own YouTube channel to benefit from it. The functionality analyzes AI-generated content to detect visual matches with the face of a registered participant. Users can then choose to request the removal of the video for privacy policy violations, submit a takedown request for copyright infringement, or take no action. However, YouTube clarifies that it will not remove all content, as parody and satire videos are permitted under its rules.

Towards Audio Support

Looking to the future, YouTube also announced that its similarity detection technology will soon include audio support. This development is part of a broader effort by the company to strengthen protections against AI-generated content. At the same time, YouTube is supporting the NO FAKES Act in Washington D.C., legislation aimed at regulating the use of AI to create unauthorized reproductions of individuals' voices and visual appearances.

Limited Impact So Far

Although YouTube has not yet disclosed the exact number of AI-generated deepfakes removed using this tool, the company indicated in March that the volume of removals remains "very low." This suggests that, while the tool is in place, its impact is still limited, but it nonetheless represents an important step in the fight against digital identity theft.