Digital news outlets urge Second Circuit to revive ChatGPT copyright claims

MANHATTAN (CN) — The digital news sites Raw Story Media and AlterNet Media asked a New York federal appeals court on Wednesday morning to overturn the dismissal of their copyright claims against ChatGPT parent OpenAI over the use of tens of thousands of copyright-protected articles to train the AI chatbot.

The two digital news outlets jointly sued OpenAI in the Southern District of New York in 2024 over what they say was the unauthorized use of their online journalism in training ChatGPT’s artificial intelligence language model dataset and the subsequent removal of copyright management information, which they claimed were violations of the Digital Copyright Millennium Act.

U.S. District Judge Colleen McMahon dismissed Raw Story and AlterNet’s complaint in November 2024 for lack of standing, holding that under the Supreme Court’s TransUniondecision, the digital publishers likely could not prove the altered works were also disseminated to the public.

On appeal before the Second Circuit, the pair of digital publishers argued McMahon, a Bill Clinton appointee, wrongly imposed the dissemination requirement for establishing standing on their copyright claims.

“Because OpenAI’s DMCA violations for removal of copyright management information closely relate to copyright infringement both at a per se level and as a result of the specific way OpenAI went about it, and because those violations also closely relate to unjust enrichment — none of which requires the dissemination the district court required — this court should reverse and permit digital publishers their day in court, just as all three of the other district courts to address the issue have done on similar pleading,” Raw Story Media and AlterNet wrote in their appeal brief.

They argue if standing under the DMCA did require dissemination, though, the lower court should have allowed jurisdictional discovery.

“OpenAI, and OpenAI alone, possesses evidence of CMI-less regurgitations, and there is a reasonable likelihood that discovery would have proven fruitful,” Raw Story and AlterNet wrote, referring to their claims of OpenAI stripping copyright management information, protected by the Digital Millennium Copyright Act, from articles it scraped online to train the ChatGPT chatbot. “Indeed, this precise discovery is being negotiated in all the other cases in the MDL.”

Attorney Steven Match told the three-judge appeals panel on Wednesday that the publisher did not mount a dissemination claim because, by the time they amended their complaint, OpenAI had changed their ChatGPT training processes in response to the many pending copyright suits against the company, which now prevent the type of “regurgitation” of copyrighted material that was relevant to their initial claims.

Match argued that Raw Story’s chief argument against OpenAI are consistent with the copyright standing of nonprofit news site The Intercept, which have been affirmed by U.S. District Judge Jed Rakoff in the same district.

“Concealing an infringement, in turn, prevents the creator from discovering it and thus enforcing its rights, and also enables the concealer to profit from the creator’s work by encouraging third parties to interact with the infringing copy rather than the original,” the publishers wrote in their brief.

U.S. Circuit Judge Dennis Jacobs, a George H.W. Bush appointee, asked Match if OpenAI was continuing to strip the copyright management information from the plaintiffs’ articles as part of the artificial intelligence training.

Match said he could not say for sure whether the CMI removal was ongoing with the latest models of ChatGPT.

“So your complaint is largely retrospective damages, unjust enrichment, stuff like that,” Jacobs responded.

OpenAI, represented at oral argument on Wednesday by attorney Charles McCloud, argued it was “entirely speculative” whether Raw Story and AlterNet would come up with evidence of regurgitations or disseminations.

“They have not alleged that there was actual copyright infringement because they do not have enforceable copyrights to begin with,” McCloud told the panel.

U.S. Circuit Judge Richard Wesley grilled McCloud on how reproduction of copyrighted material alone is not enough for standing on copyright claims.

“People who hold copyrights don’t always know if the works have been reproduced, and yet it’s quite clear that reproduction alone is a violation of the copyrights, inherent part of copyright. When you go to Chat GPT and you ask ‘how is a copyright infringed,’ The first thing it says is ‘reproducing,’” the George W. Bush appointee said.

Jacobs and Wesley were joined on the panel by U.S. Circuit Judge Eunice Lee, a Joe Biden appointee. They did not indicate how or when they would rule.

Follow @jruss_jruss