Obtaining consent for all training data used for AI models is impossible.

From Bloomberg: 2024-04-04 21:27:00

OpenAI Inc.’s intellectual property attorneys stated that obtaining consent for all copyrighted content used to train AI models would be impossible due to the large volume of data involved. Negotiating for every piece of content needed would prevent these models from existing, they added at the Vanderbilt University Music Law Summit. This fact presents a challenge for discussions on what policies are appropriate.



Read more at Bloomberg: OpenAI Attorneys Say Licensing All Training Data Is Impossible