Authors’ lawsuit cites messages on book data and job losses

A group of authors has filed new court papers accusing OpenAI and Microsoft of knowingly using copyrighted books without permission to train AI systems. The documents include internal messages in which employees discussed using LibGen, a library of pirated books, and the possibility that AI writing tools could take work from authors.

The Authors Guild publicized the filings on September 21. They were submitted in Alter v. OpenAI and Microsoft, a class-action case in Manhattan brought on behalf of the Guild and other book copyright owners. The claims are allegations in an ongoing lawsuit, not findings by a judge.

What do the filings say?

The plaintiffs cite a 2020 message from OpenAI policy director Jack Clark predicting that better GPT models could substitute for some writers’ labor. He also wrote that the work “will make people unemployed,” according to the brief.

The papers say Microsoft executives learned about OpenAI’s use of LibGen in April 2019, during a presentation by Sam Altman and Dario Amodei. They also quote OpenAI researcher Sam McCandlish describing concern about the “optics” of using material from a “sketchy Russian website.”

The filings further point to 2022 messages about removing LibGen files from OpenAI systems and storage. The Authors Guild says those messages show an effort to hide evidence; the documents alone do not establish that interpretation as a court finding. The plaintiffs also cite an OpenAI employee’s comments about having GPT models finish George R. R. Martin’s A Song of Ice and Fire series.

What happens next?

The authors are asking the court for partial summary judgment, a decision on some issues without a full trial. More written arguments are expected over the next few months, and the Guild says a hearing is expected in early 2027. The filings are one part of broader copyright litigation over whether AI companies can train on books and other works without permission.