YouTube Protects Celebrities from AI Deepfakes

⚡

Key Takeaways

1YouTube is expanding its AI deepfake detection feature to include celebrities, allowing users to report and potentially remove such content.

2The program, initially tested with creators, now includes politicians and journalists, and is open to stars without a YouTube account.

3Participants must provide an ID and a video selfie to use the tool, which focuses on facial recognition.

💡Why it matters — This initiative aims to protect the image of celebrities in light of the rise of deepfakes, while raising questions about the future monetization of such content.

YouTube Intensifies Its Fight Against AI Deepfakes for Celebrities

YouTube recently announced the expansion of its AI deepfake monitoring feature to the Hollywood industry. This decision means that certain AI-generated videos of celebrities could soon be removed from the platform. The feature, called likeness detection, is designed to search for and report AI deepfake content concerning public figures enrolled in the program. These individuals have the option to track AI content related to them on YouTube and request its removal. However, these requests are subject to YouTube's privacy policy evaluations, and not all will necessarily be accepted.

An Expanding Program

Initially, YouTube began testing this feature with content creators last fall. In March, the program was expanded to include politicians and journalists. Now, YouTube claims that this tool will be accessible to celebrities, whether or not they have an account on the platform. This opening aims to provide broader protection against deepfakes, which can harm the public image of the individuals involved.

Registration Process and Limitations

To participate in this program, celebrities must submit an ID along with a selfie video. Likeness detection specifically focuses on faces, not voices or other identifiable characteristics. It is important to note that the removal of deepfakes is not guaranteed. Certain uses, such as parody or satire, are protected and cannot be removed. YouTube indicated that content creators who have utilized this feature have requested the removal of a "very small" number of videos.

Comparison with Content ID

YouTube compares likeness detection to its Content ID system, which allows for the search and removal of copyrighted material. However, a major difference lies in the fact that, unlike Content ID, likeness detection does not yet allow rights holders to monetize videos from other users using their material. Nevertheless, it seems the industry is moving toward this monetization possibility.

New Developments and Implications

Earlier this month, YouTube introduced a new feature allowing creators to digitally clone their likeness using AI to insert it into videos. The talent agency CAA, which has supported the expansion of likeness detection, has a database containing the biometric data of its clients. This data can be used for commercial purposes. For example, TikTok star Khaby Lame sold the rights to his likeness to promote products online, although this deal has faced hurdles and has not yet been finalized according to Business Insider.

Industry Outlook

In an interview with The Hollywood Reporter, some talent agents see the rise of AI deepfakes as an opportunity for the entertainment industry to engage with fans. Some celebrities may wish to remove AI content related to them when eligible, while others may encourage fan-created AI content. In the future, artists might even accept AI deepfakes of themselves, provided they are compensated for their use.