YouTube expands secret AI acquisitions to politicians, government officials, and journalists
YouTube is expanding its match-detection technology, which identifies AI-generated deepfakes, to a test group of government officials, political candidates and journalists, the company announced Tuesday. Members of the test team will get access to a tool that detects unauthorized content generated by AI and allows them to request that it be removed if they believe it violates YouTube’s policy.
The technology itself was introduced last year to nearly 4 million YouTube creators in the YouTube partner program, following earlier tests.
Similar to YouTube’s existing Content ID system, which detects copyright-protected material in videos uploaded by users, the match detection feature looks for simulated faces created by AI tools. These tools are sometimes used to try to spread misinformation and use the way people perceive the truth, as they raise well-known people – such as politicians or other government officials – to say and do things in these AI videos that they did not do in real life.
With the new test program, YouTube aims to balance the convenience of users with the risks associated with AI technology that could produce a satisfactory likeness of a public figure.
“This expansion is about the integrity of the public conversation,” said Leslie Miller, YouTube’s Vice President of Government Affairs and Public Policy, at a press conference ahead of Tuesday’s launch. “We know that the risk of AI impersonation is particularly high for those in the public space. But while we provide this new shield, we are also careful how we use it,” he noted.
Miller explained that not all matches found will be removed upon request. Instead, YouTube will review each request under existing privacy policy guidelines to determine whether the content is satirical or political criticism, which are protected forms of free speech.
The company noted that it is advocating for these protections at the federal level, as well, through its support for the NO-FAKES Act in DC, which would regulate the use of AI to create unauthorized re-creations of a person’s voice and visual likeness.
To use the new tool, eligible test takers must first prove their identity by uploading a selfie and government ID. They can then create a profile, view matches that appear, and voluntarily request to be removed. YouTube says it plans to eventually give people the ability to block uploads of infringing content before it goes live or, perhaps, allow them to monetize those videos, similar to how its Content ID system works.
The company would not confirm which politicians or officials would be among its first testers, but said the goal is to make the technology more widely available over time.
These AI videos will be labeled as such, but the placement of these labels is inconsistent. For some, the label appears in the video’s description, while videos that focus on more “sensitive topics” will include the label at the front of the video. This is the same approach YouTube takes with all AI-generated content.
“There’s a lot of AI-generated content, but that difference isn’t really tangible in the content itself,” explained Amjad Hanif, YouTube’s Vice President of Creator Products, about the label’s placement. “It could be an AI-generated cartoon. And so I think there’s a decision as to whether it’s a possible category that deserves visual denial,” he said.
YouTube is not yet sharing how much of the removal of these types of AI deepfakes is attributable to this deep-sensing technology in the hands of creators, but noted that the amount of content removed so far is “very small.”
“I think a lot [creators]it was just to raise awareness of what is being created, but the volume of removal requests is actually really low because most of them seem inhumane or add to their overall business,” said Hanif.
That may not be the case with deepfakes of government officials, politicians, or journalists.
Over time, YouTube aims to bring its deep recognition technology to more areas, including visual speech and other creative things like popular characters.
