The New York Times report also alleges that Google modified its privacy policy in June 2022 to cover its use of publicly available content, including Google Docs and Google Sheets, for AI training. Google, however, maintains that it only uses videos from creators who have agreed to participate in an experimental program and that it only uses publicly available content with the permission of users who opt into Google's experimental features. Both OpenAI and Google have been contacted for comments on these allegations.
Key takeaways:
- OpenAI and Google have been accused of potentially violating creators' copyrights by training their AI models on text transcribed from YouTube videos, according to a report by The New York Times.
- OpenAI reportedly used its Whisper speech recognition tool to transcribe over one million hours of YouTube videos to train its GPT-4 AI model.
- Google allegedly knew about OpenAI's practices but did not take action because it was also using YouTube videos to train its own AI models, but claims it only does so with videos from creators who have agreed to participate in an experimental program.
- The report also alleges that Google updated its privacy policy in June 2022 to cover its use of publicly available content, including Google Docs and Google Sheets, to train its AI models and products, but Google insists this is only done with the permission of users who opt into its experimental features.