(Raw Story Media Inc., et al. v. OpenAI Inc., et al., No. 24-1514, S.D. N.Y.)
(Decision and order available. Document #46-241113-086Z.)
“Let us be clear about what is really at stake here. The alleged injury for which Plaintiffs truly seek redress is not the exclusion of CMI from Defendants' training sets, but rather Defendants' use of Plaintiffs' articles to develop ChatGPT without compensation to Plaintiffs. . . . Whether or not that type of injury satisfies the injury-in-fact requirement, it is not the type of harm that has been ‘elevated’ by Section 1202(b )(i) of the DMCA [Digital Millennium Copyright Act]. . . . Whether there is another statute or legal theory that does elevate this type of harm remains to be seen. But that question is not before the Court today,” U.S. Judge Colleen McMahon of the Southern District of New York said.
Raw Story Media Inc. and Alternet Media Inc. filed suit on Feb. 28 in the District Court claiming that OpenAI Inc., OpenAI GP LLC, OpenAI LLC, OpenAI OPCO LLC, OpenAI Global LLC, OAI Corp and OpenAI Holdings LLC used data scraped from “most of the internet” to train their AI, ChatGPT.
‘Nearly Verbatim’
ChatGPT will sometimes produce “nearly verbatim” works that are subject to copyright, the plaintiffs alleged. This includes times that it will “regurgitate verbatim or nearly verbatim copyright-protected works of journalism.” If a user asks ChatGPT for current events, it will at times “mimic significant amounts of material from copyright-protected works of journalism.” To the extent that ChatGPT was trained on journalism articles, it can recreate this information unless the defendants have specifically trained it to do otherwise. It will do so without crediting the author or title and without copyright notice or disclaimer, the plaintiffs said.
The plaintiffs alleges that the defendants knew that users would broadcast ChatGPT results, at least in part because the defendants advertise and market ChatGPT as a tool to generate content that can then be put in front of additional audiences.
Raw Story and Alternet alleged violation of Section 1202(b)(1) of the DMCA, 17 U.S.C. § 1202(b)(1).
The plaintiffs sought statutory damages or the total of their damages and the defendants’ profits and an injunction requiring the defendants to remove all copies of works for which author, title, copyright or terms of use information was removed.
The defendants moved to dismiss in April, arguing that the plaintiffs lacked standing or failed to state a claim. The plaintiffs’ opposition was filed May 13, and the reply on May 20. The plaintiffs moved to file a surreply, which the court denied.
Injury
Judge McMahon granted the motion to dismiss and denied the plaintiffs’ request to replead without prejudice to a proper filing that includes an explanation “why the proposed amendment would not be futile.” Judge McMahon said she agreed that the plaintiffs lacked article III, U.S. Const., art. III, standing to pursue either the claim for damages or injunctive relief.
“I am not convinced that injury for interference with property provides the necessary ‘close historical or common-law analogue’ to Plaintiffs' alleged injury. For one thing, Plaintiffs are wrong that Section 1202 grant[s] the copyright owner the sole prerogative to decide how future iterations of the work may differ from the version the owner published. Other provisions of the Copyright Act afford such protections but not Section 1202. . . . Section 1202 protects copyright owners from specified interferences with the integrity of a work's CMI. In other words, Defendants may, absent permission, reproduce or even create derivatives of Plaintiffs' works — without incurring liability under Section 1202 — as long as Defendants keep Plaintiffs’ CMI intact,” Judge McMahon said.
The DMCA’s legislative history shows it was enacted to bring the United States in line with international copyright law and not to guard against property-based injury, Judge McMahon said.
Judge McMahon said she was not convinced simply removing CMI from a work without then disseminating the CMI-less work had a historical or common-law analogue. The alleged injury must be concrete and not abstract. The plaintiffs have not shown any adverse effects from the defendants’ alleged removal of CMI from the works and resulting use as training material for AI. The plaintiffs’ argument is akin to the dissent in TransUnion LLC v. Ramirez, 594 U.S. 413 (2021), but the majority in that case rejected the position, Judge McMahon said.
The injunctive relief claim fares no better even though the defendants concede a historical and common-law analogue exists for this type of injury, Judge McMahon said. But plaintiffs do not allege that there is a risk of future reproduction of the copyrighted works by ChatGPT. Given the amount of data on which ChatGPT was trained and how it works the chances that it would reproduce copyrighted works exactly appears remote. And plaintiffs cannot allege that the information in the copyrighted works itself is protected content, Judge McMahon said.
Counsel
The plaintiffs are represented by Jonathan Loevy, Michael Kanovitz, Lauren Carbajal, Stephen Stich Match and Matthew Topic of Loevy & Loevy in Chicago.
The defendants are represented by Joseph C. Gratz and Vera Ranieri of Morrison & Foerster LLP in San Francisco and Allyson R. Bennett and Rose S. Lee of its Los Angeles office.
(Additional documents available: Memo endorsement denying surreply. Document #46-241113-087R. Plaintiffs’ motion for surreply. Document #46-240605-026B. Defendants’ memorandum in support of motion to dismiss. Document #46-240605-002B. Plaintiffs’ opposition. Document #46-240605-024B. Defendants’ reply. Document #46-240605-025B. Raw Story, et al.’s complaint. Document #46-240306-052C.)