Google is once again in controversy, this time over allegations that it used YouTube creators’ videos without explicit permission to train its Veo 3 video generation AI model. The development has raised serious questions about data usage transparency, AI ethics, and content ownership on one of the world’s largest video platforms.
What’s the controversy?
According to recent reports, Google has used over 20 billion YouTube videos to train its newly unveiled Veo 3 AI model—a system designed to convert text prompts into ultra-realistic videos. Veo 3 was introduced at the Google I/O 2025 event, and it is claimed to be Google’s most advanced video generation tool to date.
The controversy started when several YouTube creators discovered their content had been used for AI training without direct notification or consent. Many called it another case of unauthorised data harvesting, sparking debates in the creator and tech community.
What is Google Veo 3?
Veo 3 is Google’s latest AI-powered video generation model, capable of producing high-resolution, lifelike videos using simple text prompts. Its applications could range from film production and advertising to content creation and education. However, its powerful performance relies heavily on massive training datasets—most of which, in this case, allegedly came from YouTube.
What does Google say?
Responding to the allegations, Google clarified that it had entered into content licensing agreements with select media companies and creators. The tech giant stated: “Our product improvements, even in the AI era, are done while respecting agreements. Creators can opt out of having their content used for training.”
Google added that YouTube’s Terms of Service already mention that content may be used globally to improve user experiences and develop new features. A blog post from September 2024 had reportedly informed creators about these policies.
YouTube terms and creator rights
YouTube’s current Terms of Service give Google a broad license to use uploaded content worldwide to enhance products and services. Creators, however, still hold copyright over their content and can restrict its use by enabling certain content protection settings, including blocking third-party AI training from companies like Amazon, Nvidia, and Apple.