Southern District of New York (SDNY), April 2025 — A federal judge has greenlit core components of a landmark copyright infringement lawsuit brought by The New York Times, Daily News LP, and the Center for Investigative Reporting (CIR) against AI pioneers OpenAI and its strategic partner Microsoft Corporation. The decision, handed down by Judge Sidney H. Stein, denies the defendants’ attempts to dismiss the most consequential copyright-related allegations, setting the stage for what could become a defining legal battle over the future of generative AI and media rights.
What the Ruling Means
The 70-page opinion issued on April 4 affirms that the plaintiffs’ claims of direct and contributory copyright infringement—including allegations that OpenAI used millions of journalistic articles without authorization to train its language models—can proceed to the discovery phase.
Among the AI models implicated are GPT-2, GPT-3, GPT-3.5, and GPT-4, all developed by OpenAI and integrated into products like ChatGPT and Microsoft Copilot, raising wide-ranging concerns across the publishing industry about the unauthorized reuse of original reporting.
Key Legal Arguments and Court Findings
OpenAI attempted to dismiss older claims on statute-of-limitations grounds, asserting that any actions prior to December 2020 should be barred under the Copyright Act’s three-year rule. However, Judge Stein sided with The New York Times Company, concluding that the plaintiffs were not on “inquiry notice” of AI training practices until years later, especially as OpenAI operated largely without transparency during its early model development.
Furthermore, the judge rejected the tech firms’ reliance on the Sony Betamax doctrine, which argues that tools capable of substantial non-infringing uses should not be deemed illegal solely due to infringing ones. Stein ruled that such defenses are premature at this stage, especially given concrete examples of GPT outputs closely mirroring Times articles, often with misleading attributions.
Trademark and DMCA Claims: A Mixed Bag
While the ruling retains trademark dilution claims under federal and state laws, it pares back some other claims. CIR and Daily News successfully retained parts of their Digital Millennium Copyright Act (DMCA) allegations, asserting that OpenAI removed or obfuscated copyright management metadata, knowingly enabling downstream infringement.
However, the DMCA claims brought by The New York Times were dismissed due to insufficient evidence of intent. Microsoft, meanwhile, temporarily escaped all DMCA charges, but could still face exposure as discovery unfolds.
What Didn’t Survive
Two key arguments were dismissed outright:
- “Hot news” misappropriation, a rarely invoked doctrine under New York common law, was deemed inapplicable. The court found that OpenAI’s outputs lacked the time-sensitive competitive edge required under this legal theory.
- “Abridgement” claims brought by CIR, which alleged that AI-generated summaries were infringing abridged versions of original articles, were thrown out for not meeting the “substantial similarity” threshold.
Media vs. AI: Stakes Are High
This lawsuit is one of several high-profile cases testing how copyright law intersects with generative AI. Similar actions have been filed by The Authors Guild, Getty Images, and individual authors like Sarah Silverman and Paul Tremblay, each questioning whether LLMs like those from OpenAI and Anthropic can legally ingest and output protected material.
Legal experts, including James Grimmelmann of Cornell Tech, view Judge Stein’s ruling as a turning point, legitimizing media companies’ concerns and clearing the way for broader industry challenges.
What’s Next?
With discovery now imminent, OpenAI and Microsoft will be required to disclose extensive documentation about their training processes, including dataset sources, content licensing (or lack thereof), and internal communications regarding copyright concerns.
The tech companies are expected to lean heavily on the fair use doctrine, a pivotal legal shield in the U.S. copyright system. But as Judge Stein notably refrained from addressing fair use at this early stage, that battle will likely define the next phase of litigation.
The cases are:
- The New York Times Company v. Microsoft Corp., et al., 1:23-cv-11195 (SDNY)
- Daily News LP v. Microsoft Corp., et al., 24-cv-3285
- Center for Investigative Reporting v. Microsoft Corp., et al., 24-cv-4872
Industry Implications
This case could reshape how AI models are trained and monetized. If successful, media organizations could secure licensing revenue or impose legal constraints on AI developers. For tech companies, it threatens the open training model paradigm, potentially compelling them to source only licensed or public domain material—an outcome that could significantly increase development costs.
Leave a comment