YouTube now requires creators to disclose AI-generated 'realistic' content

YouTube's AI disclosure labels will start appearing across all its surfaces and formats in the coming weeks

YouTube said on March 18 that creators will now have to disclose to viewers when realistic-looking content is made using artificial intelligence (AI) tools, as part of the company's efforts to provide more transparency around AI-created content that could mislead viewers.

Starting today, YouTube is introducing a new tool in its Creator Studio that will require creators to disclose to viewers when content that a viewer can easily mistake for a real person, place, or event, was created with altered or synthetic media, including generative AI.

YouTube had first announced this move in November 2023, even as the Google-owned video platform rolls out its own AI products for creators on the platform.

The move also comes ahead of a busy election season in India and several other countries this year. In January, YouTube CEO Neal Mohan had told Moneycontrol that the platform will be "incredibly vigilant" in combating misinformation and deepfakes.

Creators will have to disclose meaningfully altered or synthetically generated content that seems realistic while uploading videos (Image source: YouTube)

Some examples of content that will require disclosures from creators are videos that use the likeness of a realistic person, wherein creators have digitally altered the content to replace the face of one individual with another or synthetically generated a person’s voice to narrate a video.

Videos that alter footage of real events or places, such as making it seem as though a real building has caught fire, or that alter a real cityscape to make it appear different than in reality will also require similar disclosures from creators. It will also apply to videos that show realistic depictions of fictional major events, such as a tornado moving toward a real town.

On the other hand, YouTube said that creators will not have to disclose content that is "clearly unrealistic" or animated, such as someone riding a unicorn through a fantastical world.

It also said that videos that include special effects such as background blur, vintage effects, color adjustment, beauty or lighting filters, and other visual enhancements will not need disclosures.

Read: Google's Gemini to curb responses to some poll-related queries ahead of Lok Sabha elections

Labels in video description panels

These disclosures will appear as a label in the description panel of the video, indicating viewers that some of the content was altered or synthetic.

For videos pertaining to more sensitive topics such as health, news, elections, or finance, YouTube will display a more prominent label on the video itself, the company said.

Examples of labels in the YouTube video's expanded description and video player

These labels will start appearing across all YouTube surfaces and formats in the coming weeks, starting with the platform's mobile apps, followed by desktop and television apps.

"While we want to give our community time to adjust to the new process and features, in the future we’ll look at enforcement measures for creators who consistently choose not to disclose this information" the company said in a blogpost.

These measures may include removal of content, suspension from the YouTube Partner Program, or other penalties. In some cases, YouTube may add a label itself even when a creator hasn't disclosed it, especially if the altered or synthetic content has the potential to confuse or mislead people, the firm said.

In September 2023, YouTube unveiled a suite of generative AI products and tools targeted at creators.

This includes the ability to create AI-generated video or image backgrounds for their short videos by just typing an idea into a prompt, helping creators come up with new video ideas, find music tracks for their videos and reach new audiences through features such as automatic dubbing.

Content created from these tools will be automatically labelled as altered or synthetic, the company had said at the time.

That said, creators using generative AI for production assistance such as generating scripts, content ideas, or automatic captions will not have to disclose content, YouTube said in a blogpost on March 18.

The company said it is also working towards an updated privacy process for consumers that will enable them to request the removal of AI-generated or other synthetic or altered content that simulates an identifiable individual, including their face or voice.

The system, first announced in November, is expected to be introduced across the world in the forthcoming future.