YouTube is set to broaden its AI-driven likeness detection technology, specifically targeting deepfake content involving government officials, political candidates, and journalists. This initiative, announced by the company on Tuesday, will enable a select group of participants to utilize a tool designed to identify unauthorized AI-generated videos and request their removal if they contravene YouTube's policies.
The likeness detection feature, which initially launched last year for about 4 million creators in the YouTube Partner Program, aims to combat misinformation by identifying AI-generated simulations of public figures. These deepfakes can mislead viewers by portraying individuals saying or doing things they never actually did.
Through this pilot program, YouTube seeks to strike a balance between protecting user expression and addressing the potential risks posed by AI technology. Leslie Miller, YouTube's Vice President of Government Affairs and Public Policy, emphasized the importance of maintaining the integrity of public discourse, particularly for those in civic roles. She remarked, "The risks of AI impersonation are particularly high for those in the civic space."
While not every detected deepfake will be automatically removed, YouTube will assess each request based on its privacy policies, distinguishing between content that constitutes parody or political critique--both of which are safeguarded forms of expression.
In addition to this new tool, YouTube is advocating for federal regulations through its support of the NO FAKES Act, which aims to govern the unauthorized use of AI to replicate individuals' voices and likenesses.
To participate in the pilot, eligible users must verify their identities by submitting a selfie and a government-issued ID. Once verified, they can create a profile, review detected matches, and request removals. YouTube envisions expanding this technology to allow individuals to prevent the upload of violating content before it goes live, similar to its existing Content ID system.
Although the identities of the initial pilot testers remain undisclosed, YouTube aims to eventually make this technology widely accessible. The platform plans to label AI-generated videos, although the placement of these labels may vary depending on the content's sensitivity.
Amjad Hanif, YouTube's Vice President of Creator Products, noted that while a significant amount of AI-generated content exists, the distinction regarding its labeling is essential for clarity. He stated, "There's a lot of content that's produced with AI, but that distinction's actually not material to the content itself."
Looking ahead, YouTube intends to expand its deepfake detection capabilities further, potentially incorporating recognizable voices and other forms of intellectual property.
