Google Using YouTube Videos to Train Gemini AI Sparks IP Dispute

Google has confirmed it uses YouTube creators’ videos to train its Gemini and Veo 3 AI models, drawing criticism over intellectual property rights. While YouTube’s policy allows video use for research and product improvement, creators say they never consented to AI training and cannot opt out once content is published. Google insists it only uses a small portion of its 20 billion-video library and will honor existing agreements with creators and media companies. The company has added indemnification clauses in Veo 3 and partnered with Creative Artists Agency to protect top creators’ image rights. However, experts warn even minimal use could involve billions of minutes of content and lead to unfair competition, potential legal disputes, and risks of AI-generated deepfakes replacing original work.