Google is using YouTube videos to train Veo 3 and creators can’t opt out
Google is using YouTube videos to train its Veo 3 AI model, and CNBC says creators can’t opt out from having their uploads included.
Google launched Veo 3, its newest AI video generative model, back in May 2025, and it has taken social media by storm thanks to how realistic it’s becoming.
Many have wondered how they’ve managed to make it so good, andCNBC saysit’s because Google is using videos onYouTubeto train the AI model.
Creators cant opt-out from training
Not only is Google using YouTube videos to train its Veo 3 audio and video generator, uploads are being used to help advance its ChatGPT competitor, Gemini, as well.
Google confirmed the news to CNBC, but says it doesn’t use every video available on the platform.
“We’ve always used YouTube content to make our products better, and this hasn’t changed with the advent of AI,” said the company’s spokesperson.
“We also recognize the need for guardrails, which is why we’ve invested in robust protections that allow creators to protect their image and likeness in the AI era — something we’re committed to continuing.”
YouTube shared back in September that content may be used to “improve the product experience… including through machine learning and AI applications.” However, the publication says creators have no way to turn this off.
Back in December, the company did make a rather-large update tohow videos are used for AI training when it comes to third party models.
Through the dashboard, creators can choose whether they let companies scrape their videos, and can even pick which ones they want to share things with. Google’s Gemini is notably missing from the list, however, which makes sense because it’s a first-party model.
American YouTube users fight back as AI analyzes watch history & demands ID
Experts warn against YouTube’s “creepy” AI age estimation system launching in the US
Markiplier responds after getting called out for using AI
