Encyclopaedia Britannica and its subsidiary Merriam-Webster filed a lawsuit in Manhattan federal district court against OpenAI, accusing the company of illegally using its articles and dictionary definitions to train and run language models.
“ChatGPT is built on freebies that defendants took from reliable, high-quality content created by the dedicated work of researchers, authors, editors and other creative people,” the lawsuit states.
Britannica and Merriam-Webster point out in the lawsuit that ChatGPT is diverting traffic that previously went to their resources, causing them direct economic damage.
According to Britannica, OpenAI copied almost 100 thousand online encyclopedia articles to train ChatGPT. The lawsuit notes that “in response to user requests, ChatGPT generates responses that copy or imitate plaintiffs’ content—sometimes verbatim.”
A separate claim concerns the model’s hallucinations: according to the plaintiffs, OpenAI infringes their trademarks when ChatGPT generates fictitious content and attributes it to Britannica or Merriam-Webster, and also reproduces incomplete and inaccurate copies of materials next to the plaintiffs’ registered trademarks.