• by Ray Schultz , Columnist, November 10, 2024

OpenAI won a signal legal victory last week, with a federal court ruling that its motion to dismiss a lawsuit filed against it by Raw Story Media and Alternet Media, Inc. is “GRANTED in its entirety.” At the same time, plaintiffs’ motion  to re-plead “is denied with prejudice."

According to the plaintiffs, ChatGPT "gives the impression that it is an all-knowing 'intelligent' source of the information being provided." 

"ChatGPT does not have any independent knowledge of the information provided in its response,” the complaint continues. “Rather, ChatGPT is trained on large amounts of text, known as ‘training sets.’ These training sets range from collections of links posted on the website Reddit to a scrape of most of the internet." 

The plaintiffs allege that "thousands" of their copyrighted works of journalism were caught in this "scrape," stripped of their author, title, and copyright information, and input into at least three of OpenAI's training sets (WebText, WebText2, and Common Crawl). Plaintiffs allege that these three training sets were then used to train ChatGPT.

The judge continues that since ChatGPT was not provided with the author, title, and copyright information, Plaintiffs allege that ChatGPT would not have learned to communicate that information when fashioning responses to user inquiries that are based on their copyrighted works, and that indeed ChatGPT "generally does not provide the author, title, and copyright information applicable to the works on which its responses are based." 

He adds that defendants may, absent permission, reproduce or even create derivatives of Plaintiffs' works - without incurring liability under Section 1202 – as long as Defendants keep Plaintiffs' CMI intact.