YouTube hands creators more power over AI model training by other companies like OpenAI, Meta, Amazon and more

YouTube has announced a significant update that will allow creators to control third-party AI access to their content for training large language models (LLMs). This move comes as AI companies increasingly seek high-quality, human-created data to improve their models, particularly in the growing field of video generation AI.

The new feature will let creators decide which AI firms can use their videos for training, while also ensuring that unauthorised scraping remains banned under YouTube’s Terms of Service.

YouTube will roll out the feature in the coming days through a new option in Studio Settings under a section titled “Third-party training”. Creators can:

    •    Grant or deny access to specific AI companies.
    •    Allow all listed companies or only select ones to use their content for training purposes.

Companies currently included on the list are OpenAI, Meta, Amazon, Adobe, Apple, Microsoft, Anthropic, xAI, Nvidia, IBM, and others.

YouTube emphasised that only content approved by creators or applicable rights holders will be eligible for AI training. The platform reiterated its strict anti-scraping policies, ensuring companies cannot bypass these settings to gather content illegitimately.

As AI companies race to source data for training models, the demand for large-scale, high-quality content has intensified. Many firms have turned to publicly available posts and content-partnership agreements. For example, Grok uses public posts on X, while Meta AI is trained on Facebook and Instagram data.

However, YouTube—home to vast amounts of human-generated video content—has been cautious about unauthorised use, aiming to protect creators’ intellectual property. With the rise of AI tools capable of generating videos, such data has become even more valuable.

YouTube’s new settings strike a balance between supporting AI innovation and respecting creators’ rights.

Notably, the announcement did not address whether AI firms would compensate creators for the use of their videos. However, YouTube stated that it remains committed to exploring collaborative opportunities between creators and third-party companies.

This update follows YouTube’s earlier efforts to protect creators from deepfakes and unauthorised imitations. The platform has been rolling out tools that give creators greater control over how their content is used, including features to detect and report fake likenesses in videos.