OpenAI, Microsoft Urge Judge to Toss The Intercept’s DMCA Suit

A Manhattan federal judge pushed back on OpenAI Inc.‘s and Microsoft Corp.‘s arguments that he should dismiss Intercept Media Inc.'s digital copyright suit at a hearing Friday.

Judge Jed S. Rakoff of the US District Court for the Southern District of New York appeared to struggle with Microsoft’s argument that The Intercept’s claims under the 1998 Digital Millennium Copyright Act should be dismissed because the publication failed to allege a proper claim and didn’t show an actual or threatened injury.

The Intercept alleged that OpenAI and Microsoft stripped copyright management information—such as authors’ names and story titles—from articles it used to train its AI models, in violation of the DMCA. Pulling up the statute, Rakoff said, “The injury is that you remove that information knowing or having reason to know it would lead to copyright infringement. Why isn’t that enough?”

Microsoft’s attorney, Annette Hurst from Orrick, Herrington & Sutcliffe LLP, responded that any injury required showing a scheme to infringe, as well as the public dissemination of the CMI-stripped information.

“I’m having problems as to why this is a legal issue versus a factual one,” Rakoff said.

Rakoff asked The Intercept’s counsel why it hadn’t alleged copyright infringement, instead taking the DMCA route.

Matt Topic from Loevy & Loevy answered that The Intercept didn’t have copyright registrations for its articles, as would be required to seek infringement damages—but it didn’t need to with its theory of liability. The tech giants copied the publication’s work by removing the CMI, Topic said, and under the DMCA, “injury in fact is the copy.”

But removing the CMI didn’t create a situation where someone using an AI model could pull copyrighted information, Joseph C. Gratz from Morrison & Foerster argued for OpenAI. Although The Intercept’s amended complaint added instances of OpenAI’s ChatGPT spitting out near-verbatim snippets of the publication’s articles, Gratz argued the examples were weak.

Pointing to one exhibit showing a side-by-side comparison of an Intercept article next to an AI output, Gratz said the article goes on for another 600 words, but the AI prompt didn’t share that information. Rakoff noticed the third example of regurgitation provided by The Intercept showed an even shorter copied sample in the AI output.

Microsoft’s April motion to dismiss said similar CMI claims have been axed in other lawsuits, calling The Intercept’s case the “skimpiest of the lot.” Microsoft said The Intercept didn’t identify any AI outputs that contained that information.

The Intercept amended its complaint in June, adding examples of verbatim or near-verbatim regurgitations of copyrighted articles and thousands of URLs from its website that it said were contained in one of OpenAI’s training datasets.

Rakoff said he plans to issue a ruling by Nov. 22.

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

The case is The Intercept Media, Inc. v. OpenAI, Inc., S.D.N.Y., No. 1:24-cv-01515, oral argument held 11/1/24.