The biggest current news in copyright AI lawsuits is a $1.5 billion class action settlement with Anthropic, a wave of federal rulings splitting on whether training AI on copyrighted works is fair use, and active cases against OpenAI, Meta, Midjourney, and Stability AI moving toward decisive rulings in 2026. No single legal theory has won out. Courts agree that training is transformative, but they’ve divided sharply on market harm and on what happens when the training data came from pirate sites.
Anthropic’s $1.5 Billion Settlement
The largest resolution to date came in Bartz et al. v. Anthropic PBC, a class action brought by authors Andrea Bartz, Kirk Wallace Johnson, and Charles Graeber. They alleged Anthropic downloaded hundreds of thousands of books from the pirate sites Library Genesis and Pirate Library Mirror to train its Claude chatbot.1Susman Godfrey. Susman Godfrey Secures $1.5 Billion Settlement in Landmark AI Piracy Case
Before the deal, Judge William Alsup of the Northern District of California issued a ruling on June 23, 2025 that has become the reference point for every AI copyright case since. Training AI models on copyrighted books is “exceedingly transformative” and qualifies as fair use, he held. But keeping pirated copies in a permanent digital library is not, because those copies “plainly displaced demand for Authors’ books — copy for copy.”2AFS Law. Landmark Ruling AI Copyright Fair Use vs. Infringement Bartz v. Anthropic
The settlement covers roughly 482,000 books and pays an estimated $3,000 to $3,100 per eligible work. Anthropic must destroy the original pirated files and certify whether they were used in commercially released models. The deal grants no license for future training.3Copyright Alliance. Participating in the Bartz v. Anthropic Settlement Preliminary approval came on September 25, 2025. A final approval hearing was held on June 3, 2026, and the motion is under submission with Judge Araceli Martínez-Olguín. About 93 percent of the class, covering 448,000 works, submitted claims; only 350 opted out.4Courthouse News Service. Authors Publishers Near Final Approval of $1.5 Billion Anthropic Copyright Settlement
The OpenAI Multidistrict Litigation
The broadest litigation against any AI company is In re OpenAI, Inc. Copyright Infringement Litigation, consolidating twelve separate lawsuits in the Southern District of New York before Judge Sidney H. Stein. Plaintiffs include The New York Times, the Chicago Tribune, and numerous individual authors, all alleging OpenAI used their work to train ChatGPT without permission.5Norton Rose Fulbright. AI in Litigation Series: An Update on AI Copyright Cases in 2026
The MDL has already produced significant procedural rulings. In October 2025, the court denied OpenAI’s motion to dismiss, finding plaintiffs had adequately alleged some ChatGPT outputs could be substantially similar to their copyrighted works. In January 2026, Judge Stein ordered OpenAI to produce 20 million de-identified ChatGPT interaction logs. In March 2026, the court compelled an additional 78 million logs to help plaintiffs assess whether ChatGPT functions as a market substitute for their content.5Norton Rose Fulbright. AI in Litigation Series: An Update on AI Copyright Cases in 2026
The New York Times case within the MDL is the highest-profile piece. The Times is seeking billions of dollars in statutory and actual damages and has asked for the destruction of GPT models trained on its content. Summary judgment briefing closed in April 2026, with a ruling expected in the third quarter. If any claims survive, trial could take place in late 2026 or 2027.6AI Lawsuit Tracker. New York Times v. OpenAI
Gracenote Media Services separately filed a database-related infringement suit against OpenAI on March 10, 2026, which has since been folded into the MDL.7CourtListener. Gracenote Media Services LLC v. OpenAI Foundation
Meta’s Two-Front Fight
Meta is defending two significant cases over its Llama models, and each tests a different theory.
In Kadrey v. Meta Platforms, Inc., Judge Vince Chhabria granted Meta summary judgment on the core training claim on June 25, 2025, ruling the use was “highly transformative.” He cautioned, however, that in many future cases, copying protected works without permission could still be illegal.8Jones Day. Two US Courts Address Fair Use in GenAI Training Cases The plaintiffs then pivoted. Their new theory: when Meta downloaded books via BitTorrent, it simultaneously uploaded those files back onto the network, seeding pirated copies and helping others infringe. In March 2026, Judge Chhabria “reluctantly” granted the plaintiffs leave to add a contributory infringement claim on that basis. Meta argues the plaintiffs cannot show it knowingly induced anyone else’s infringement, and the case remains at the summary judgment stage.9Ars Technica. Meta Hopes SCOTUS Piracy Ruling Will Help It Beat Lawsuit Over Torrenting AI Data
The second case is Elsevier Inc. et al. v. Meta Platforms, Inc. and Mark Zuckerberg, filed May 5, 2026 in the Southern District of New York. Elsevier, Cengage, Hachette, Macmillan, and McGraw Hill brought the class action alongside novelist Scott Turow, who serves as class representative for authors.10Publishers Weekly. Publishers File Infringement Lawsuit Against Meta, Zuckerberg The complaint alleges Meta used “millions of textual works” from pirate sites to train Llama, names Zuckerberg as personally authorizing the infringement, and cites internal communications indicating Meta considered raising its dataset licensing budget to $200 million before Zuckerberg shut the effort down. The publishers say the reason he killed the plan was that licensing even a single book would undermine the company’s fair use defense.11New York Times. Publishers File Infringement Lawsuit Against Meta, Zuckerberg Meta has said it “will fight this lawsuit aggressively.”
Hollywood Studios v. Midjourney
In June 2025, Disney, Universal, DreamWorks, Marvel, Lucasfilm, and Twentieth Century Fox sued Midjourney in the Central District of California. It was the first copyright lawsuit by major Hollywood studios against an AI company. Their 110-page complaint called Midjourney a “quintessential copyright free-rider” and alleged the platform can generate recognizable images of Yoda, Darth Vader, and Marvel characters without any specialized prompt engineering.12Georgetown Law Tech Institute. Disney, NBC Universal, and DreamWorks File Major IP Lawsuit Against AI Image Generator Midjourney
In November 2025, the court consolidated that action with a separate suit filed by Warner Bros., DC Comics, Hanna-Barbera, Turner Entertainment, and the Cartoon Network. The parties estimate a 14-day trial and are referred to a private mediator, with mediation required by August 2026. Expert discovery deadlines run through late 2026, with dispositive motions expected after that.13ForensisGroup. Disney and Universal v. Midjourney: U.S. Generative AI Copyright Litigation Over Image Training and Outputs
Notably, Disney is simultaneously licensing on the other side of the industry. In December 2025, Disney announced a three-year deal allowing OpenAI’s Sora platform to generate short videos featuring more than 200 Disney, Pixar, Marvel, and Star Wars characters, along with a $1 billion equity investment in OpenAI. The deal excludes actor likenesses and voices and includes restrictions on how characters can be depicted. Disney’s general counsel drew the contrast: “In this deal, OpenAI is both respecting and valuing our creativity.”14Wall Street Journal. Disney to Invest $1 Billion in OpenAI, License Characters for Use in ChatGPT, Sora
Other Active Cases to Know
Thomson Reuters v. Ross Intelligence
This case produced one of the first judicial findings that AI training infringes copyright. In February 2025, Judge Stephanos Bibas ruled that Ross Intelligence’s use of 2,243 Westlaw headnotes to train its legal search tool was not fair use, finding the product was commercially competitive and not transformative. The decision is on interlocutory appeal to the Third Circuit, and Ross Intelligence has ceased operations, citing the financial burden of the litigation.15Authors Alliance. Thomson Reuters v. Ross: The First AI Fair Use Ruling Fails to Persuade
Andersen v. Stability AI
A class action by visual artists challenging the use of the LAION image dataset to train AI image generators, this case names Stability AI, Midjourney, and DeviantArt as defendants. Trial is set to begin September 8, 2026 before Judge William Orrick in the Northern District of California, which would make it the first AI copyright case to reach a jury.16NYU JIPEL. Andersen v. Stability AI: The Landmark Case Unpacking the Copyright Risks of AI Image Generators
Concord Music Group v. Anthropic
Universal Music Group, Concord, and ABKCO are suing Anthropic over Claude’s ability to reproduce copyrighted lyrics from at least 500 songs, including works by Beyoncé, the Rolling Stones, and the Beach Boys.17Reuters. US Music Publishers Suing Anthropic Make Their Case Against AI Fair Use The court denied the publishers’ preliminary injunction motion in March 2025 as “overbroad,” though a stipulated order requires Anthropic to maintain guardrails against outputting infringing lyrics. The publishers asked Judge Eumi Lee in March 2026 to rule before trial that Anthropic infringed and that fair use does not apply.18Loeb & Loeb. Concord Music Group Inc. v. Anthropic PBC
Doe v. GitHub (The Copilot Case)
Software developers are suing GitHub, Microsoft, and OpenAI over allegations that the Copilot coding tool stripped copyright management information from open-source code in violation of the DMCA. The case is stayed at the district level while the Ninth Circuit considers an interlocutory appeal. Oral arguments were held on February 11, 2026, and a decision is pending.19Baker McKenzie. The Copilot Litigation
Music AI Class Actions
Independent songwriters and recording artists have filed separate class actions against AI music generators: against Suno and Udio in October 2025, against Kunlun Tech and Skywork AI (behind Mureka) in December 2025, and against Google over its Lyria 3 and ProducerAI tools in March 2026.20Loevy + Loevy. Music AI Class Action
Where Courts Have Split on Fair Use
The central legal question — whether training AI on copyrighted material is fair use — has not produced a unified answer. Rulings so far share a common thread on the first fair use factor (purpose and character of the use), but diverge on market harm and on the significance of pirated versus lawfully obtained training data.
In both Bartz and Kadrey, courts found that training AI models is “highly” or “exceedingly” transformative because models learn patterns from text rather than reproducing it. Judge Alsup compared the process to how humans read and learn. In Thomson Reuters v. Ross Intelligence, however, the court reached the opposite conclusion, finding the use was commercial, not transformative, and directly competitive with the original product.21Reed Smith. Court AI Fair Use Thomson Reuters Enterprise GmbH v. Ross Intelligence
Market harm has emerged as the swing factor. Judge Chhabria ruled for Meta in Kadrey in part because the plaintiffs failed to present evidence that Meta’s specific actions caused actual harm to the market for their particular books, even while acknowledging AI-generated content could “dramatically undermine the incentive for human beings to create.”8Jones Day. Two US Courts Address Fair Use in GenAI Training Cases Judge Alsup, meanwhile, found clear market harm where Anthropic kept pirated copies, because those copies displaced sales “copy for copy.” The provenance of training data, whether purchased or pirated, is shaping up as one of the most consequential distinctions in AI copyright law.22Authors Alliance. Anthropic Wins on Fair Use for Training Its LLMs, Loses on Building a Central Library of Pirated Books
What to Watch in the Rest of 2026
Several milestones will shape how these questions get resolved. The Andersen v. Stability AI trial is scheduled for September 2026. The OpenAI MDL expects a summary judgment ruling in the third quarter. The Ninth Circuit’s decision in the Copilot DMCA appeal could redefine the obligations of AI companies that train on open-source code. And the Thomson Reuters appeal at the Third Circuit may produce the first appellate ruling on fair use in the AI training context.
New cases keep arriving. Seven AI-related copyright suits were filed in the first three months of 2026 alone, targeting companies from Runway AI to Adobe to Snap.23Copyright Alliance. Artificial Intelligence Copyright Court Cases With statutory damages of up to $150,000 per infringed work and training datasets spanning millions of copyrighted works, the financial stakes for AI companies remain enormous, and the legal framework for resolving them is still being written one ruling at a time.