OpenAI, Authors Get Rules Set for Inspecting AI Training Data

OpenAI Inc. for the first time will reveal training data for its artificial intelligence models in authors’ consolidated lawsuit accusing the company of using their copyrighted works train ChatGPT, per rules in a federal judge’s order.

Magistrate Judge Robert M. Illman signed off on the proposed inspection protocol outlining the process for the plaintiffs’ team to review OpenAI’s training data in an order filed Tuesday in the US District Court for the Northern District of California.

The protocol stipulates all information reviewed will be tagged highly confidential and for attorneys’ eyes only, according to the order. Plaintiffs’ experts reviewing the ...