Intercept’s Evidence Enough to Advance OpenAI Claim, Judge Says

The Intercept provided enough specific examples of how its articles were stripped of details like author names to support its claims OpenAI Inc. violated the Digital Millennium Copyright Act by training on its works, a federal judge said Thursday.

The publication plausibly alleged OpenAI intentionally removed “copyright management information” from its articles in violation of the DMCA, Judge Jed S. Rakoff wrote in an opinion explaining his previous decision to let the publication’s lawsuit move forward in the US District Court for the Southern District of New York.

He pointed to several pieces of evidence, including the Intercept’s identification of specific training sets it claimed OpenAI used to train ChatGPT and specific URLs from its web domain that its data scientist mined from approximations of the datasets as evidence for the assertions, as well as evidence of how algorithms like “Dragnet” and “Newspaper” can remove information besides the main text of its articles.

“Of course, these allegations rely in part, even if implicitly, on information and belief,” he wrote. “But the court does not expect more at this early stage of the litigation, particularly because of OpenAI’s secrecy over the contents of the sets used to train the latest versions of ChatGPT.”

Rakoff denied OpenAI’s motion to dismiss Intercept Media Inc.'s lawsuit in November and pledged to release a full opinion explaining the reasoning. His opinion Thursday said the publication had shown OpenAI “intentionally” stripped details like author names from articles plugged into training datasets, “knowing” it would facilitate infringement.

The November order marked the Intercept case as one that would pave the way for other generative AI-related lawsuits alleging a violation of the DMCA—which prohibits the removal of CMI like article titles and author names—to survive the motion to dismiss stage.

The Intercept included evidence of regurgitation and examples of how the Dragnet and Newspaper algorithms worked in an amended complaint filed in June last year. The publication originally sued OpenAI and Microsoft Corp. in February.

Beyond plausibly alleging intentionality, the Intercept needed to show OpenAI removed CMI knowing it would “induce, enable, facilitate or conceal” copyright infringement, Rakoff said. He wrote he was persuaded by the Intercept’s theory of “downstream infringement” by ChatGPT users, because the AI model was billed as tool that can “generate content for a further audience.”

He also noted OpenAI’s indemnification policy for “commercial” users sued for copyright infringement “provides additional support for its allegation that the company knew some users would likely infringe copyright.”

Matt Topic, Loevy & Loevy attorney representing the Intercept, said they were pleased “the court confirmed our core claims against OpenAI can go forward.”

“We look forward towards the ultimate resolution of the case which has important ramifications for all copyright owners,” Topic said.

OpenAI’s attorneys didn’t immediately respond to a request for comment.

Rakoff’s November order dismissed the Intercept’s claims against Microsoft. Those claims lacked the same “factual specificity” as the claims against OpenAI, he wrote Thursday.

Rakoff also said the Intercept’s allegations that Microsoft and OpenAI shared copies of the Intercept’s works with each other, violating the DMCA, draws support “exclusively from general observations about the business relationship between” the two companies.

Microsoft’s attorneys also didn’t immediately respond to a request for comment.

Latham & Watkins LLP, Morrison & Foerster LLP, and Keker & Van Nest LLP represent OpenAI. Orrick, Herrington & Sutcliffe LLP represents Microsoft.

The case is The Intercept Media Inc. v. OpenAI Inc., S.D.N.Y., No. 1:24-cv-01515, opinion filed 2/20/25.