USA TODAY Co., the US newspaper group formerly known as Gannett, and 13 affiliated newspaper companies sued the OpenAI Foundation and six other OpenAI entities in the US District Court for the Southern District of New York on 8 October (case no. 1:26-cv-08892), alleging that OpenAI scraped and used news content from 19 of their publications without permission to train its GPT models and that ChatGPT outputs copy or rework their reporting. The complaint says OpenAI's models copied hundreds of thousands of articles and other materials from these publications; the plaintiffs seek more than USD 250 million in damages and a permanent injunction, and ask for the destruction, under section 503(b) of the US Copyright Act, of GPT and other large language models and training sets that incorporate their content.
This is a joint suit by several publishing entities, not a class action; the case is at the complaint stage, the complaint sets out the plaintiffs' allegations, and the court has made no findings.
Compared with earlier suits of this kind, the complaint devotes considerable space to the commercial market for news content. First, a licensing market already exists: the complaint points out that OpenAI has signed content licensing agreements with more than a dozen news and media organisations, including the Associated Press, Axel Springer, The Atlantic and Vox Media, to show that using news works requires permission and payment.
Second, the plaintiffs' own licensing business is harmed: they license content to other media and publishers for a fee, and the complaint says OpenAI undermines these arrangements by providing their content directly to readers. Third, substitution: citing OpenAI internal documents and statements by executives, the complaint says its products are largely substitutive for news, and that ChatGPT receives about 1 million prompts a week on local news.
If OpenAI raises a fair use defence, these facts go to the fourth fair use factor, the effect of the use on the potential market for or value of the work; whether a licensing market has formed bears directly on how that factor is weighed, although courts have not reached a settled view in AI training cases.
J&C notes that content owners planning to enforce copyright in the United States should be aware that every work listed by the plaintiffs carries a US copyright registration, and that claims for statutory damages and attorneys' fees generally depend on timely registration; entering into content licences and keeping records of permissions and fees may also later serve as evidence that a licensing market exists.
Source: www.courtlistener.com | storage.courtlistener.com
