A report by The New York Times claims both OpenAI and Google transcribed videos from YouTube and used these texts to train their AI models, possibly violating copyrights. OpenAI reportedly transcribed more than one million hours of YouTube videos.