AI copyright lawsuits have piled up faster than courts can resolve them. Authors, news publishers, visual artists, record labels, and software developers have all sued the companies behind ChatGPT, Claude, Llama, Midjourney, Stability AI, Suno, and Udio, arguing that training generative models on copyrighted works without permission is infringement. Federal courts issued their first substantive rulings on the fair use question in 2025 and reached different answers. One case produced a $1.5 billion class settlement. Several of the largest suits remain years from trial.
The Consolidated Cases Against OpenAI and Microsoft
By early 2025, the volume of copyright suits against OpenAI and Microsoft was large enough that the Judicial Panel on Multidistrict Litigation consolidated twelve of them into a single proceeding in the Southern District of New York before Judge Sidney H. Stein. The plaintiffs include The New York Times, the Authors Guild, the Center for Investigative Reporting, Raw Story Media, The Intercept, and groups of individual authors.1McKool Smith. MDL Panel Consolidates Twelve Generative AI Copyright Cases
In October 2025, Judge Stein denied the defendants’ motion to dismiss the direct copyright infringement claims, holding that the plaintiffs had adequately alleged OpenAI copied their text and that a reasonable jury could find AI-generated outputs substantially similar to the originals.2Loeb & Loeb LLP. In Re OpenAI Inc Copyright Infringement Litigation Analysts do not expect the consolidated case to conclude in 2026.3AI Business. AI Lawsuits: Settlements, Licensing Deals, Litigation
The most prominent piece of the MDL is the Times lawsuit, filed in December 2023, accusing OpenAI and Microsoft of copying millions of articles to train large language models that now compete with the paper’s publishing business.3AI Business. AI Lawsuits: Settlements, Licensing Deals, Litigation On April 4, 2025, Judge Stein denied OpenAI’s statute-of-limitations argument on direct infringement, allowed contributory infringement and certain DMCA claims to proceed, and dismissed common-law unfair competition claims with prejudice.4Justia. The New York Times Company v. Microsoft Corporation No trial date has been set. The Authors Guild class action, also inside the MDL, is in class certification proceedings.5Baker & Hostetler LLP. Case Tracker: Artificial Intelligence Copyrights and Class Actions
Fair Use Rulings Have Split
The core question in almost every AI copyright case is whether ingesting protected works to train a model qualifies as fair use. Three federal rulings in the first half of 2025 came out differently.
Thomson Reuters v. Ross Intelligence
The first federal court to decide the question ruled against the AI company. On February 11, 2025, Judge Stephanos Bibas found that Ross Intelligence infringed Thomson Reuters’ copyrights by using 2,243 Westlaw headnotes to train a competing legal research tool.6Justia. Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc. The court rejected the fair use defense, finding the use commercial and non-transformative because Ross built a direct competitor to Westlaw, and it recognized a potential market for AI training data that Thomson Reuters has the right to control. Ross has since shut down.7Authors Alliance. Thomson Reuters v. Ross: The First AI Fair Use Ruling Fails to Persuade Ross’s tool was a non-generative search product, so commentators have questioned how far the ruling reaches for generative AI companies.
Bartz v. Anthropic
On June 23, 2025, Judge William Alsup in the Northern District of California reached the opposite conclusion for a generative model. He ruled that Anthropic’s use of lawfully purchased books to train its Claude language models was “spectacularly” transformative fair use, because the models extract statistical patterns rather than reproducing the original works, and he found no evidence of market harm. But he drew a hard line at pirated material. Anthropic’s acquisition and storage of books from pirate sites like LibGen was “inherently, irredeemably infringing” and not fair use, because the pirated copies displaced demand for the authors’ books “copy for copy.” What Anthropic did with the pirated library beyond training is headed to trial.8Authors Alliance. Anthropic Wins on Fair Use for Training Its LLMs, Loses on Building a Central Library of Pirated Books
Kadrey v. Meta
Two days later, on June 25, 2025, Judge Vince Chhabria granted summary judgment to Meta, finding that training Llama on copyrighted books was “highly transformative” fair use.9Justia. Kadrey et al v. Meta Platforms Inc. The court acknowledged AI-generated content could theoretically flood the market and harm authors but ruled for Meta because the plaintiffs failed to produce evidence that Meta’s specific models had caused or were likely to cause that harm. Judge Chhabria emphasized the ruling did not broadly bless Meta’s training practices and suggested plaintiffs with stronger market-effect evidence could win in the future.10Authors Alliance. Meta Wins on Fair Use for Now, but Court Leaves Door Open for Market Dilution
The $1.5 Billion Anthropic Settlement
The largest financial resolution so far came in Bartz v. Anthropic itself, separate from the fair use ruling on training. The parties reached a $1.5 billion class action settlement covering Anthropic’s use of pirated books from the LibGen and PiLiMi datasets. The class covers copyright owners of books that appeared in those datasets, had an ISBN or ASIN, and were registered with the U.S. Copyright Office within five years of publication.11Authors Guild. What Authors Need to Know About the Anthropic Settlement
About 500,000 titles qualify, and each is expected to receive at least $3,000, split between authors and publishers. For trade and university press titles, the default is a 50/50 split when both parties file claims.11Authors Guild. What Authors Need to Know About the Anthropic Settlement Anthropic will pay the fund in four installments through September 2027.12ClassAction.org. Bartz et al v. Anthropic PBC Notice
Visual Artists Sue Image Generators
In Andersen v. Stability AI, visual artists Sarah Andersen, Kelly McKernan, and Karla Ortiz filed a class action in early 2023 against Stability AI, Midjourney, DeviantArt, and Runway AI, alleging billions of images were scraped to train image generators. On August 12, 2024, Judge William Orrick dismissed the DMCA claims with prejudice but allowed direct copyright infringement, induced infringement, and trademark claims to proceed, accepting the theory that AI models may contain copyrighted works as “algorithmic or mathematical representations.”13Copyright Alliance. Andersen v. Stability AI Copyright Case The case remains active.
Getty Images’ separate suit against Stability AI in Delaware, over more than 12 million photographs, was voluntarily dismissed without prejudice after Getty said it would refile in the Northern District of California.14Baker & Hostetler LLP. Getty Images v. Stability AI
Music Labels: Settlements, Licenses, and One Holdout
The three major record labels sued AI music generators Suno and Udio in June 2024, alleging training on copyrighted recordings without permission.15RIAA. Record Companies Bring Landmark Cases for Responsible AI Against Suno and Udio Two of the three labels have now settled.
Universal Music settled with Udio on October 29, 2025, agreeing on compensation and license terms for a joint subscription music platform launching in 2026, with UMG artists paid for both training and outputs on an opt-in basis.16Universal Music Group. Universal Music Group and Udio Announce Strategic Agreements Warner Music settled with Suno on November 25, 2025, in a deal that included Warner’s sale of the Songkick concert-discovery platform to Suno and a licensing partnership under which Suno will launch new licensed models in 2026, replacing its existing ones. Warner artists retain control over whether their names, likenesses, voices, and compositions are used.17TechCrunch. Warner Music Signs Deal With AI Music Startup Suno, Settles Lawsuit
Sony Music has not settled with either company and is the sole major label still litigating against both Suno and Udio. A fair use ruling in the Udio case is expected in summer 2026 and could set a significant precedent for AI music.18Chartlex. Music Industry AI Lawsuits Tracker
Music publishers Universal Music Publishing Group, Concord, and ABKCO are separately pursuing Anthropic over Claude’s use of song lyrics. The court denied Anthropic’s motion to dismiss claims for contributory and vicarious infringement and for removal of copyright management information, and plaintiffs have sought a preliminary injunction requiring guardrails against infringing lyric outputs.19Baker & Hostetler LLP. Concord Music Group Inc. v. Anthropic PBC
The GitHub Copilot Case Has Narrowed Sharply
Programmer and lawyer Matthew Butterick filed the Copilot case in November 2022 against GitHub, Microsoft, and OpenAI over the use of open-source code to train AI coding tools. Judge Jon Tigar dismissed most of the original 22 claims, including the core allegation that Copilot generates code identical to copyrighted work, leaving only two breach-of-contract counts.20Legal.io. Judge Throws Out Majority of Claims in GitHub Copilot Lawsuit Plaintiffs appealed the DMCA dismissal to the Ninth Circuit, which held oral argument on February 11, 2026, and a decision is pending.21Baker & Hostetler LLP. The Copilot Litigation
News Publishers Target Perplexity
Dow Jones and NYP Holdings, publishers of The Wall Street Journal and the New York Post, sued Perplexity AI in October 2024 on a distinct theory: that Perplexity’s retrieval-augmented system copies articles into its database and repackages them into user-facing responses containing verbatim reproductions, sometimes with fabricated text falsely attributed to the publishers. On August 21, 2025, the court denied Perplexity’s motion to dismiss in full, finding the company subject to New York jurisdiction and allowing the copyright and trademark claims to proceed.22FindLaw. Dow Jones and Company Inc. v. Perplexity AI Inc.
Can AI-Generated Work Be Copyrighted?
A parallel question is whether AI outputs can receive copyright protection at all. In Thaler v. Perlmutter, Stephen Thaler sought to register an image created entirely by his AI system. The D.C. Circuit affirmed the Copyright Office’s denial on March 18, 2025, holding that the Copyright Act requires authorship “in the first instance by a human being.”23U.S. Court of Appeals, D.C. Circuit. Thaler v. Perlmutter, No. 23-5233 The Supreme Court declined to hear the case on March 2, 2026, making the ruling final.24Baker Donelson. Supreme Court Denies Certiorari in Thaler v. Perlmutter
How much human involvement is required for AI-assisted work to qualify is still open. In Allen v. Perlmutter, pending in Colorado, artist Jason Allen is challenging the Copyright Office’s refusal to register his award-winning image made with Midjourney, arguing that using an AI tool is comparable to using a camera.25Law360. Allen v. Perlmutter
Product Liability Lawsuits Over Chatbot Harm
A separate category of AI litigation frames chatbots as defective products rather than protected speech. The leading case is Garcia v. Character Technologies, filed after the 2024 suicide of 14-year-old Sewell Setzer III, whose mother alleges the teenager was harmed by prolonged interactions with a Character.AI chatbot. On May 21, 2025, Judge Anne Conway allowed most claims to proceed and ruled that the Character.AI chatbot is a “product” for product liability purposes, not a service protected by the First Amendment.26Transparency Coalition. Important Early Ruling in Character.AI Case: This Chatbot Is a Product, Not Speech The court also found the plaintiff plausibly alleged Google is liable as a component-part manufacturer for supplying the underlying language model.27FindLaw. Megan Garcia v. Character Technologies Inc.
Raine v. OpenAI, filed in 2025, alleges ChatGPT contributed to a teenager’s self-harm through the absence of crisis-intervention guardrails. A March 2026 class action against xAI alleges Grok was used to generate child sexual abuse material from real photographs of minors, with researchers estimating Grok produced roughly three million sexualized images in an eleven-day period spanning late 2025 and early 2026.28Lieff Cabraser Heimann & Bernstein. AI Deepfakes
Federal Regulators and the Apple Intelligence Settlement
On September 25, 2024, the Federal Trade Commission launched Operation AI Comply, a coordinated sweep against companies using AI to deceive consumers. DoNotPay settled for $193,000 over claims its “AI Lawyer” could substitute for human attorneys without evidence the tool had been tested or staffed by legal professionals. Ascend Ecom was accused of defrauding consumers of at least $25 million with “AI-powered” storefront promises, and a federal court halted the operation. Rytr was charged with enabling subscribers to generate fake consumer reviews at scale.29FTC. FTC Announces Crackdown on Deceptive AI Claims and Schemes Evolv Technologies settled in March 2025 over allegedly false claims about its AI-powered security screening. The FTC has continued through 2026, banning operators of AI business-opportunity schemes and opening inquiries into AI companion chatbots marketed to children.30FTC. Artificial Intelligence
In a consumer case unrelated to training data, Apple agreed in May 2026 to pay $250 million to settle a class action alleging it misled iPhone buyers about “Apple Intelligence” features and an enhanced Siri. The complaint claimed Apple promoted AI features “that did not exist at the time, do not exist now, and will not exist for two or more years, if ever.”31BBC. Apple Intelligence Settlement The class covers U.S. consumers who bought qualifying iPhone 15 and iPhone 16 models between June 2024 and March 2025, with eligible claimants receiving between $25 and $95 per device. Apple did not admit wrongdoing.32Mashable. Apple Class Action Settlement iPhone Models
State Laws Add Another Layer
State legislatures moved in parallel with the courts. Tennessee’s ELVIS Act, signed in March 2024, protects musicians from unauthorized AI-generated voice replicas. California’s governor signed 18 AI-related bills on September 29, 2024, including protections for performers’ digital likenesses and training-disclosure requirements for AI developers, with key provisions taking effect in 2026. Illinois banned unauthorized use of AI-generated digital likenesses, and more than half of U.S. states had enacted laws targeting malicious deepfakes by the end of 2025.33American Bar Association. Generative AI Copyright Law: Current Trends34WilmerHale. Year in Review: Artificial Intelligence Privacy Litigation Trends
Courts have not agreed on whether training generative models on copyrighted works is fair use. Legislatures are still experimenting. Several of the largest suits are years from resolution. The rulings and settlements so far have begun to set the outer edges of what AI companies can do with other people’s work, but the central question of who owes what for the data that powers modern AI has not been answered.