During the 2025 Google I/O developer conference, Google introduced Veo 3. This innovation has been recognized as the most advanced AI video-generating system to date. It can create cinematic, "entertainment-quality" videos that include both audio and dialogue, marking a significant advancement in the field of generative AI.
According to reports from CNBC, the model was trained on an extensive collection of YouTube videos, which has raised concerns among content creators.
Google allegedly took a subset of the twenty billion videos on YouTube to train Veo 3 and its Gemini AI models. While the YouTube terms of service does permit such usage, the majority of the creators are unaware of the consequences. Creators can block the use of their content by companies like Apple or Nvidia for AI.
Unfortunately, there is no such opt-out for Google. This has frustrated the issues of consent, ownership, and transparency even more in the training of AI.
The creators grant YouTube a royalty-free, sublicensable license pursuant to Section 3 of the Terms of Service relating to uploading content to YouTube. It was stated in a 2024 blog post that content might be used for "improving product experience," which includes machine learning and AI applications.
This very broad clause has certainly enabled Google to incorporate user-generated videos into its AI tools.
Many independent creators and media companies complain that their works are being used to train AI systems that will eventually compete with them or replace them. Recently, Disney and Universal sued Midjourney for copyright infringement. The entertainment giants stated that the company monetized AI-generated content derived from their iconic characters.
Midjourney filed the V1 model soon after, which was the first-ever text-to-video model, sparking another round of the fair-use debate in AI development.
Google's use of YouTube content to train Veo 3 has raised questions about the legality of AI training methods. As generative tools like Veo 3 and Midjourney V1 continue to evolve, there may be a need for clearer regulations and transparent consent frameworks. This would help strike a balance between innovation and the rights of content creators.