• On Tuesday, the YouTube platform announced a series of measures around Artificial Intelligence.
  • In the coming months, content creators will have to inform their subscribers if a piece of content has been modified in a synthetic way. If they don't, their videos could be removed and penalties could take place.
  • However, YouTubers will be better protected from the risks of deepfakes with stronger moderation.

Will the YouTube of tomorrow be more transparent when it comes to Artificial Intelligence (AI)? At least that's the promise made by the platform on Tuesday in a press release. "YouTube's vision is to transform the platform experience for creators and their audiences through generative AI. This transformation must be done by striking a balance between opportunities and a responsibility to protect the YouTube community," the announcement summarizes. For the video platform, Artificial Intelligence functions as much as a technological asset and a danger for its creators.

Let's recall, for example, the pornographic deepfakes that Léna Situations was the victim of this summer. During her vlogs in August, the YouTuber was worried – rightly – about the use of her face on another body, naked, for pornographic purposes. Musical artists have also been affected, such as the fake duo Drake-The Weeknd, who have never actually released a song together. Less dangerous but just as fake, the famous Squeezie had seen his face changed to that of Emmanuel Macron. There are thousands of examples like these.

Manipulated content removed

To combat this proliferation, YouTube promises to combat these images "digitally generated without permission or to misrepresent a point of view." To this end, Internet users will soon be able to "submit a privacy claim to remove manipulated, synthetically created or AI-generated content that simulates an identifiable person (including with their face or voice)," YouTube announces. However, not all videos can be deleted. Artificial intelligence will be able to detect more fake content around a public figure and some videos could be maintained if they are considered satire or parody.

"Our music industry partners will also be able to request the removal of AI-generated audio content that mimics the unique voice of an artist singing or rapping," the platform adds. Different tools will then be used, in particular by looking for other occurrences through the artist's current events. Going back to the example of Drake-The Weeknd, the AI would see that no media promotion has been made about it and could flag the song as potentially fake.

Mandatory wording

On the other hand, content creators will also have to cooperate to make synthetic images more intelligible to their audience. While many rules already exist on YouTube to prohibit content modified using technical means, the platform now intends to do more. "New forms of AI storytelling can also be used to generate content that can deceive viewers, especially if viewers are unaware that the videos have been edited or created from scratch," the platform explains. In the coming months, content creators will therefore be required to indicate whether they have modified any of their content with a synthetic tool. For example, a video of a real personality speaking something different from reality. A label will then appear in the description and in the case of a more sensitive topic, may even be visible on the video player.

"This is especially important when the content deals with sensitive topics, such as elections, ongoing conflicts, public health crises, or public figures," YouTube said. If you forget, the fault could cost the content creator dearly. Its content could be removed and it could be subject to sanctions. "Sometimes language is not enough to reduce the risk of harm. Synthetically created content that violates the Community Guidelines will be removed from the platform, regardless of whether it contains a label," the platform confirms.

  • Tech
  • By the Web
  • YouTube
  • Deepfake
  • Artificial Intelligence (AI)