YouTube Pilots Deepfake Likeness Tool for U.S. Officials, Candidates and Journalists

6d Developing 1 source 4 relevant data

YouTube is expanding its AI‑driven likeness detection tool beyond entertainment creators to a pilot group of government officials, political candidates and journalists, with plans to open it to any user in those categories. Announced in a March 10 Axios interview with executives, the system scans uploaded videos for facial impersonations and allows verified participants—who must submit a government ID and video selfie—to review flagged clips and request takedowns through YouTube’s privacy complaint process, though parody and satire remain allowed. The move is framed by YouTube’s government‑affairs chief Leslie Miller as aimed at protecting the "integrity of the public conversation" at a time when generative AI has made it easier to fabricate convincing videos of public figures, including President Trump. CEO Neal Mohan has made AI transparency and synthetic‑media protections one of his top 2026 priorities, and YouTube is also backing the proposed federal NO FAKES Act while pointing to Trump’s earlier TAKE IT DOWN Act on non‑consensual intimate images as a narrower precedent. Company officials say creators using the tool so far have requested relatively few removals and often view impersonations as benign or even helpful to their business, but YouTube is now exploring voice‑impersonation detection and possible monetization models for likeness‑based content, underscoring how platform rules are racing to catch up with politically sensitive deepfake risks.

AI and Deepfakes Policy Social Media Platforms and Elections

📌 Key Facts

YouTube is extending its likeness detection tool to a select pilot group of government officials, political candidates and journalists, with future access planned for any user in those groups.
The tool scans uploaded videos for facial likenesses, then lets verified participants request removals via YouTube’s privacy complaint process, while still permitting parody and satire.
YouTube executives say the expansion is meant to protect the integrity of civic discourse amid rising AI‑driven deepfakes, and the company has endorsed the proposed federal NO FAKES Act as a blueprint for takedown requirements.
The feature, first developed in 2024 with Creative Artists Agency and tested by high‑profile creators like MrBeast and Marques Brownlee, was opened to all creators last year and now is being adapted for high‑risk civic figures.
YouTube is exploring extensions of the tool to detect voice impersonation and to allow targets to monetize content using their likeness, similar to its existing Content ID system.