Google is using YouTube to train its latest AI-enabled video generation tool, Veo 3, using its extensive video library uploaded on the platform. Google confirmed this with CNBC and said it was using only a subset rather than the entire YouTube video to train AI models.
The report added that several creators are “unconsciously concerned” about helping to train models.
YouTube allows creators to opt out of providing content for training to third-party AI companies, but it cannot prevent Google from doing so. Third-party companies such as Apple, Anthropic, Amazon, Meta, and Microsoft can use YouTube data to train generated AI models.
VEO 3 is the company's third generation AI video generation model and is the latest model. It was announced last month at Google's I/O 2025 and can generate 8-second video.
Additionally, VEO 3 can add audio to generated video. This is a feature that does not exist on some competing platforms, including Openai's SORA. Initially it was available in the US alone, but the company has since expanded access to more than 70 countries.
Google Gemini Pro Plan users are $20 per month and have limited access to the model. However, the high-end Gemini Ultra Plan, which costs $249 per month, offers the best use of the VEO 3.
