YouTube is giving creators and rights holders a clearer way to say whether third-party AI companies may train models on their videos. The new control, announced on Monday, appears inside YouTube Studio and is designed as an opt-in system rather than a default permission slip.
The change arrives after creators raised concerns that major AI companies had trained models on their work without consent or compensation. It also comes as AI video systems, including OpenAI’s Sora, have sharpened questions about how online video should be used in generative AI development.
What YouTube is changing
Starting today, creators and rights holders can use a new setting in YouTube Studio to flag whether specific third-party AI companies have permission to train models on their content. The feature is optional, and creators can choose which companies, if any, they want to authorize.
YouTube is presenting an initial list of 18 companies. The list includes AI21 Labs, Adobe, Amazon, Anthropic, Apple, ByteDance, Cohere, IBM, Meta, Microsoft, Nvidia, OpenAI, Perplexity, Pika Labs, Runway, Stability AI, and xAI.
The company says these businesses were selected because they are building generative AI models and are likely to be sensible candidates for creator partnerships. Creators are not limited only to that list, however. A broader option labeled "All third-party companies" allows any third party to train on the creator’s data, even if that company is not named in the initial menu.
The practical shift is straightforward: YouTube is making permission more visible and easier to express inside the platform where creators already manage their channels. Instead of assuming a creator is open to AI training, the new system requires a creator-side choice.
Who can use the new control
The setting is available to eligible creators with access to the YouTube Studio Content Manager and an administrator role. Those users can view or change their third-party training settings from YouTube Channel settings at any time.
YouTube says creators globally will be alerted through banner notifications in YouTube Studio on desktop and mobile over the next few days. That rollout matters because the control is only useful if creators know it exists and understand that the default position has changed.
By default, creators will not allow third parties to train on their videos. That default makes the platform’s position clearer to outside companies: unless a creator has opted in, third-party training is not authorized through this new control.
What the setting does not change
The new option is focused on third-party access. It does not alter YouTube’s Terms of Service, which already prohibits third parties from accessing creator content in unauthorized ways, including scraping.
It also does not stop Google from training its own AI models on some YouTube content. YouTube told TechCrunch that Google will continue doing so in accordance with its existing agreement with creators.
That distinction is central to understanding the update. The setting gives creators a way to permit or withhold third-party AI training, but it is not a broader rewrite of YouTube’s relationship with Google’s own AI work.
YouTube was unable to say whether the setting could have any retroactive effect on third-party AI model training that has already taken place. The company did point back to its Terms of Service, which says third parties cannot access creator content without authorization.
Why creators pushed for more control
The rise of generative AI has made creator consent a more visible conflict. YouTube creators complained that companies including Apple, Nvidia, Anthropic, OpenAI, and Google itself had trained AI models on their material without consent or compensation.
The issue is especially sensitive for video creators because their work can include images, voices, performances, editing, and recognizable styles. As AI video tools improve, the question of who gets to train on online video becomes more than a technical matter. It becomes a question of permission, attribution, and potential payment.
YouTube said this fall that it would address the issue in the near future. The Studio setting is the first concrete step described in the source article: a control that lets creators say yes to selected AI companies rather than leaving consent ambiguous.
Where this could go next
YouTube sees the feature as a first step toward making it easier for creators who want to allow AI training on their videos. The platform also frames it as a possible path toward compensation for that training.
A later phase may involve giving authorized companies access to direct downloads of creators’ videos. That would make the permission system more operational, because a selected AI company would not only be recognized as authorized but could also receive a more direct way to access the approved content.
The company has also described related AI detection plans. In September, YouTube unveiled plans for creator controls for AI training and announced new AI detection tools intended to help prevent creators, artists, musicians, actors, and athletes from having likenesses such as faces and voices copied and used in other videos.
That detection technology would expand on YouTube’s existing Content ID system, which had previously focused only on copyright-protected material. Together, the training control and detection tools show YouTube trying to address both sides of the AI video problem: who can train on creator content, and how creators can respond when their likeness is copied.
The timing also reflects the wider pace of AI video development. Separately, Google’s AI research lab DeepMind announced Veo 2 on Monday, a video-generating AI model aimed at rivaling OpenAI’s Sora.