# Raw Story v. OpenAI: Article III Standing Halts a DMCA Training Claim

> An SDNY judge dismissed publishers' DMCA claim over copyright info stripped for AI training: removal without dissemination is too abstract an injury.

Topic: Copyright  |  Author: Lidiia Levitska  |  Source: Intellectual Property Law (outsideipcounsel.com)
Canonical: https://outsideipcounsel.com/blog/raw-story-v-openai-cmi-standing-training/


Standing, the constitutional requirement that a plaintiff show a concrete injury, is usually a preliminary skirmish. In the litigation over generative artificial intelligence, it has become a decisive battleground. *Raw Story Media, Inc. v. OpenAI, Inc.*, No. 1:24-cv-01514 (S.D.N.Y.), delivered one of the first rulings to dismiss a DMCA claim against an AI developer on standing grounds. On November 7, 2024, Judge Colleen McMahon of the United States District Court for the Southern District of New York held that publishers whose articles had their author credits and other copyright-management information stripped for use in training data had not suffered a concrete injury sufficient to sue, because the stripped works were never disseminated.

The decision reframed a central question in AI copyright litigation. It is not enough for a plaintiff to allege that a developer removed copyright-management information when it ingested content to train a model. The plaintiff must also show that the removal caused a concrete, particularized harm of the kind Article III demands. *Raw Story* said that bare CMI removal, absent downstream dissemination, does not clear that bar, and in doing so it opened a split with a parallel case in the same courthouse.

## At a glance

- **Case:** *Raw Story Media, Inc. v. OpenAI, Inc.*, No. 1:24-cv-01514 (S.D.N.Y. Nov. 7, 2024).
- **Decided:** November 7, 2024; opinion by Judge Colleen McMahon dismissing the case for lack of Article III standing.
- **Holding:** Removing copyright-management information from articles used to train ChatGPT, without disseminating the stripped works, is too abstract an injury to confer standing under *TransUnion LLC v. Ramirez*.
- **Status:** Judge McMahon denied leave to amend on April 3, 2025, and Judge Sidney Stein denied reconsideration on June 18, 2025. The appeal, No. 25-1756 (2d Cir.), was argued on March 18, 2026 and, as of July 16, 2026, remained undecided. The dismissal stands in tension with the contrary approach in *The Intercept Media, Inc. v. OpenAI, Inc.*

## The statute and the standing overlay

The Digital Millennium Copyright Act, at 17 U.S.C. Section 1202(b), prohibits intentionally removing or altering copyright-management information, and distributing works knowing that such information has been removed, where the actor knows or has reasonable grounds to know that the conduct will induce, enable, facilitate, or conceal an infringement. Copyright-management information includes identifying details like the author's name, the title, and terms of use. Publishers have argued that when AI developers scrape articles and strip this information before feeding the text into training pipelines, they violate Section 1202(b).

But a statutory violation is not automatically a case a federal court may hear. Under Article III of the Constitution, a plaintiff must have standing, which requires an injury in fact that is concrete, particularized, and actual or imminent. The Supreme Court sharpened that requirement in *TransUnion LLC v. Ramirez*, 594 U.S. 413 (2021), holding that a bare statutory violation, without concrete harm, does not suffice. *Raw Story* applied *TransUnion* to the CMI context and asked whether stripping author credits from training data, standing alone, inflicts the sort of concrete injury the Constitution requires.

## The facts and posture

Raw Story Media and AlterNet Media are news organizations that have published hundreds of thousands of articles online, including investigative reporting and opinion columns. They alleged that OpenAI copied their articles to train ChatGPT and other models and, in doing so, removed the copyright-management information those articles carried, such as author bylines and title and terms-of-use data. Their complaint pleaded a violation of Section 1202(b). Notably, they did not press a direct copyright-infringement claim of the kind at issue in other AI suits; their theory rested on the DMCA's CMI provisions.

OpenAI moved to dismiss, arguing among other things that the publishers lacked standing. Judge McMahon agreed and granted the motion. The court found that the publishers had not identified a concrete injury, because the alleged CMI removal happened inside the training process and the stripped versions of their articles were not shown to have been disseminated to anyone.

## The court's reasoning: no dissemination, no concrete harm

The heart of the ruling is the link between dissemination and injury. The court reasoned that the purpose of Section 1202(b) is to protect the integrity of copyright-management information as works move through the world, so that downstream users can identify rights holders and terms. When information is stripped but the resulting works are not distributed, the feared harm, that a recipient will encounter a work shorn of its rights information, does not materialize. The injury the publishers described, that their content was used to train a model without preserved credits, was in the court's view too abstract to satisfy *TransUnion*.

The publishers' separate request for an injunction failed on a different ground. The court accepted that a forward-looking risk of harm can support standing, but concluded that because ChatGPT was trained on a scrape of most of the internet, the plaintiffs had not plausibly alleged a substantial risk that the current version would reproduce one of their articles without the required credits.

The court denied the publishers' motion for leave to replead, though only without prejudice to renewal on a proper record, meaning a proposed amended pleading accompanied by an explanation of why amendment would not be futile. Judge McMahon wrote that she was skeptical about the plaintiffs' ability to allege a cognizable injury but was prepared to consider an amended pleading at least as to injunctive relief. She later denied leave to amend on April 3, 2025, finding that the proposed amended DMCA claim still rested on the bare statutory violation of CMI removal. The ruling did not decide whether OpenAI in fact removed CMI or whether training on the articles infringed; it held only that these plaintiffs, on this theory, had not shown the concrete harm that would let a federal court reach those questions.

## A split inside the same courthouse

*Raw Story* did not produce a consensus. In a parallel Southern District of New York case, *The Intercept Media, Inc. v. OpenAI, Inc.*, Judge Jed Rakoff took a more permissive view, allowing a Section 1202(b)(1) removal claim to survive dismissal while dismissing the Section 1202(b)(3) distribution claim and dismissing the case against Microsoft in full. The divergence created a district-level split on whether and when CMI-removal claims against AI developers can proceed past the pleading stage. That split matters because the two cases arise from materially similar conduct, scraping and stripping news content for training, yet reach different conclusions about whether such conduct inflicts a cognizable injury.

The practical effect is uncertainty. Until the Second Circuit resolves the question, which it may do in the pending *Raw Story* appeal, the viability of a CMI claim against an AI developer in the Southern District of New York can depend on the judge and on how the plaintiff frames the injury. *Raw Story* stands for the proposition that CMI removal in training, without dissemination, is not enough; *The Intercept* signals that some CMI theories can advance. Plaintiffs are left to plead concrete downstream harm as specifically as they can, and defendants are left to press *Raw Story* wherever the facts allow.

## Open questions

- **Will the Second Circuit resolve the split?** The court heard argument in the *Raw Story* appeal, No. 25-1756, on March 18, 2026, and as of July 16, 2026 had not issued a decision. The answer will shape CMI litigation strategy nationwide.
- **What downstream harm suffices?** The ruling leaves unclear exactly what a plaintiff must show, perhaps reproduction of stripped content in outputs, to convert an abstract CMI-removal allegation into a concrete injury.
- **Does the analysis change for direct-infringement plaintiffs?** *Raw Story* involved a DMCA-only theory. How standing interacts with CMI claims brought alongside direct-infringement claims remains to be developed.

## Implications for creators and businesses

- **Plead concrete harm, not just removal.** Publishers and other rights holders bringing CMI claims against AI developers should identify a tangible downstream injury, such as dissemination or reproduction of the stripped works, rather than resting on the act of removal alone.
- **Standing is now a front-line defense.** AI developers facing DMCA claims should test Article III standing early, because *Raw Story* shows it can dispose of a CMI theory before the merits are ever reached.
- **Forum and framing matter.** The intra-district split means outcomes can turn on how a claim is pleaded and where it is filed, so strategy should account for the unsettled state of CMI standing law.
- **Watch for appellate movement.** Because the question is unresolved at the circuit level, both plaintiffs and defendants should monitor developments closely and preserve arguments for appeal.

## Frequently asked questions

**Why did Raw Story lose its DMCA claim against OpenAI?**
Because Judge McMahon found the publishers lacked Article III standing. Removing copyright management information like author names from articles used to train ChatGPT, without disseminating the stripped works, was too abstract an injury to be a concrete harm under the Supreme Court's TransUnion decision.

**What is the split with The Intercept v. OpenAI?**
In a parallel SDNY case, a different judge allowed a Section 1202(b)(1) CMI claim against OpenAI to proceed past dismissal, reaching a different conclusion on whether such claims can go forward. The two rulings created a district-level split on CMI standing in AI training suits.

**Does Raw Story bar all CMI claims against AI developers?**
No. It holds that CMI removal during training, without dissemination of the stripped works, does not by itself create Article III standing. A plaintiff who can show concrete downstream harm, such as reproduction of the stripped content in outputs, may still be able to establish standing.

## Authorities and sources

- [Raw Story Media, Inc. v. OpenAI, Inc., order dismissing for lack of standing (S.D.N.Y. Nov. 7, 2024) (PDF via PatentlyO)](https://patentlyo.com/media/2024/11/Raw-Story-Order.pdf)
- [Raw Story Media, Inc. v. OpenAI Inc., district court docket and filings (RECAP archive)](https://archive.org/details/gov.uscourts.nysd.616533)
- [17 U.S.C. Section 1202, integrity of copyright management information (Cornell LII)](https://www.law.cornell.edu/uscode/text/17/1202)
- [Loeb & Loeb, "Raw Story Media, Inc. v. OpenAI, Inc."](https://www.loeb.com/en/insights/publications/2024/11/raw-story-media-inc-v-openai-inc)
- [Bloomberg Law, "OpenAI Beats Raw Story Copyright, Training Lawsuit, for Now"](https://news.bloomberglaw.com/ip-law/openai-defeats-raw-story-copyright-training-lawsuit-for-now)
- [Mass Law Blog, "SDNY Courts Split Over Copyright Management Information in AI Cases"](https://masslawblog.com/copyright/sdny-courts-split-over-copyright-management-information-in-ai-cases/)