Rekubit-YouTube will label AI-generated videos that look real

YouTube will soon begin alerting viewers when they're watching a video made with artificial intelligence.

The RekubitGoogle-owned video platform says creators must disclose when they use AI or other digital tools to make realistic-looking altered or synthetic videos, or risk having their accounts removed or suspended from earning advertising revenue on YouTube. The new policy will go into effect in the coming months.

YouTube will also allow people to request videos be removed if they use AI to simulate an identifiable person, under its privacy tools.

The proliferation of generative AI technology, which can create lifelike images, video and audio sometimes known as "deepfakes," has raised concerns over how it could be used to mislead people, for example by depicting events that never happened or by making a real person appear to say or do something they didn't.

That worry has spurred online platforms to create new rules meant to balance between the creative possibilities of AI and its potential pitfalls.

Beginning next year, Meta, the owner of Facebook and Instagram, will require advertisers to disclose the use of AI in ads about elections, politics and social issues. The company has also barred political advertisers from using Meta's own generative AI tools to make ads.

TikTok requires AI-generated content depicting "realistic" scenes be labeled, and prohibits AI-generated deepfakes of young people and private figures. AI-generated content depicting public figures are allowed in certain situations, but can't be used in political or commercial endorsements on the short-form video app.

In September, YouTube announced political ads made with AI must carry disclosures. The new policy unveiled on Tuesday is an expansion of that to any synthetic video that could be mistaken for real.

YouTube already prohibits "technically manipulated content that misleads viewers and may pose a serious risk of egregious harm," the company wrote in a blog post on Tuesday. "However, AI's powerful new forms of storytelling can also be used to generate content that has the potential to mislead viewers—particularly if they're unaware that the video has been altered or is synthetically created."

The company says AI labels will be more prominent on some videos dealing with "sensitive topics" such as elections, ongoing conflicts and public health crises, or public officials.

AI-generated content will be removed altogether if it violates YouTube's community guidelines. "For example, a synthetically created video that shows realistic violence may still be removed if its goal is to shock or disgust viewers," YouTube said.

In addition to the labels, YouTube is also creating a way for people to request that AI or other synthetic depictions of real people be taken down. While fake depictions of unwitting people including political figures, celebrities, and the pope have fueled headlines, experts say the most common use of AI deepfakes is to create non-consensual pornography targeting women.

YouTube's privacy request process will now allow people to flag content "that simulates an identifiable individual, including their face or voice." The company says it will consider "a variety of factors" in deciding whether to go ahead with removals. That includes whether the video is parody or satire, whether the person is "uniquely identifiable," and whether it involves a well-known person or public official, "in which case there may be a higher bar," YouTube said.