OpenAI would have used automatic transcriptions of YouTube videos to train Sora. The same technique would have been used by Google to improve the capabilities of Gemini.
Google would have trained its AI models, from which the Gemini chatbot was also born, only on video content granted for use by the authors, specific clauses in the contracts. In any case, it would be a violation of the rules of the Alphabet platform, as highlighted in an interview with Bloomberg by NealMohan, CEO of YouTube.