Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Monday YouTube announced this will give creators more choices about how third parties can use their content to train AI models. Starting today, creators and rights holders will be able to tag YouTube if they allow specific third-party AI companies to train models on the creator’s content.
From a new setting in the creator dashboard, YouTube Studio, creators will be able to opt-in to this new feature if they choose. Here they will see a list of 18 companies that have permission to train on the creator’s videos.
Companies included in the initial list include AI21 Labs, Adobe, Amazon, Anthropic, Apple, ByteDance, Cohere, IBM, Meta, Microsoft, Nvidia, OpenAI, Perplexity, Pika Labs, Runway, Stability AI and xAI. YouTube notes that these companies were chosen because they build generative AI models and are likely smart choices to partner with creators. However, creators will also be able to choose a setting that says “All third-party companies,” meaning they allow any third-party to teach their data, even if they’re not listed.
The company also notes that eligible creators are those with access to YouTube Studio Content Manager with an administrator role. They will also be able to view or change third-party training settings on the YouTube Channel parameters at any time.
After the rise of artificial intelligence technology and especially OpenAI’s Sora-like AI videoYouTube creators Apple, Nvidia, Anthropic, OpenAI and even companies Google itself, among others, trained AI models on their materials without their consent or compensation. YouTube said this fall will solve this problem in the near future.
But while the addition of the setting will control third-party access, the company tells TechCrunch that Google will continue to train its AI models on some YouTube content in line with the existing agreement with creators. The new setting is otherwise unchanged YouTube Terms of Servicethis prohibits third parties from accessing creative content through unauthorized means, such as scrapping.
Instead, YouTube sees the feature as a first step toward making it easier for creators who want to allow companies to train AI in their videos, and perhaps as a way to get compensated for that training. In the future, YouTube will likely take this process a step further by allowing companies to allow their creators to directly download their videos.
With the introduction of the feature, the default setting for all creators will not allow third parties to practice on their videos, making it more obvious to companies that already do so, and they are doing so against the creators’ wishes.
YouTube was unable to say whether the new setting could affect any third-party AI model training in retrospect. But the company says its Terms of Service specify that third parties cannot access creative content without permission.
The company first announced plans to offer control to creators for AI training in Septemberwhen it also announced new AI detection tools to help prevent creators, artists, musicians, actors and athletes from having their likenesses, including their faces and voices, copied and used in other videos. The detection technology will expand YouTube’s existing Content ID system, which previously focused only on copyrighted material, the company explained at the time.
Creators globally will be alerted to the new feature via banner notifications in YouTube Studio on desktop and mobile over the next few days.
Separately, Google’s AI research lab DeepMind announced Veo 2, a new video-generating AI model that aims to compete with OpenAI’s Sora, on Monday.
