AI Lawsuit Wave: Anthropic Settlement, NYT v. OpenAI, Fair Use

More than 75 AI lawsuits have been filed in U.S. courts since 2022, and the rulings coming out of them are starting to define what artificial intelligence companies can and cannot do.¹ The disputes cover copyright infringement, hiring discrimination, defamation, product liability, and harm to minors. Some have already produced landmark outcomes, including a $1.5 billion class settlement from Anthropic and the first federal rulings on whether AI training qualifies as fair use. Many of the largest cases are still headed toward trial or awaiting appellate review, and several rulings expected in the second half of 2026 could reshape the industry.

Fair Use and AI Training: The Central Legal Fight

The question threading through most of the copyright suits is whether feeding copyrighted books, articles, images, songs, and code into an AI model counts as fair use. Two federal judges in the Northern District of California gave the first real answers in June 2025.

In Bartz v. Anthropic, Judge William Alsup held that Anthropic’s use of legally purchased books to train Claude was “quintessentially transformative” and qualified as fair use. The AI was not trained on the works “to race ahead and replicate or supplant them,” the court reasoned, but “to turn a hard corner and create something different.”² But Judge Alsup drew a hard line at pirated material. Books Anthropic downloaded from pirate libraries such as LibGen and PiLiMi were, in his words, “inherently, irredeemably infringing.”³

Two days later, Judge Vince Chhabria reached a similar conclusion in Kadrey v. Meta Platforms, calling Meta’s use of copyrighted books to train Llama “highly transformative.” He then narrowed his own ruling, writing that it “does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful. It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”⁴ Had the authors argued that Meta’s AI could produce works competing directly with their books, or that AI-generated content floods the market and pushes theirs out, the outcome might have been different.⁵

A third case cuts the other way. In Thomson Reuters v. ROSS Intelligence, Judge Stephanos Bibas in Delaware rejected a fair use defense outright. ROSS had used Westlaw’s copyrighted headnotes to train a competing legal search engine. The court found the use non-transformative because it served the same purpose as the original and was designed to build a market substitute.³ The Third Circuit heard oral arguments on June 11, 2026, and its forthcoming decision could be the first appellate ruling on AI training and fair use.⁶

Two patterns are emerging. Courts have been more receptive to fair use when a generative AI produces something functionally different from its training data, and more skeptical when the AI is a direct competitor to what it learned from. And the source of the training data matters: legally purchased material earns a hearing on transformativeness, while pirated material does not.

The $1.5 Billion Anthropic Settlement

The same Bartz v. Anthropic case that produced the fair use ruling also produced the largest copyright class action ever certified. Three authors alleged Anthropic downloaded copyrighted books from pirate libraries to train Claude.⁷ On July 17, 2025, Judge Alsup certified a class of copyright owners whose books appeared in the LibGen and PiLiMi datasets Anthropic had accessed. The class potentially reached up to 7 million claimants.⁸

Anthropic told the appeals court that certification was a “death knell,” with potential damages reaching hundreds of billions of dollars at the $150,000 statutory maximum per work.⁸ Instead of going to trial, the company settled for $1.5 billion.

The non-reversionary fund covers roughly 482,460 identified works, guaranteeing each qualifying work at least $3,000. The default split between authors and publishers is 50/50, and disputes over distribution go to a court-appointed special master.⁹ Anthropic pays in four installments through September 2027 and must destroy all books downloaded from the pirate datasets.¹⁰

A fairness hearing was set for May 14, 2026. By spring, about 99,450 claims had been filed covering 264,809 works, or 54% of eligible titles. Only 350 rightsholders opted out, less than 0.5% of the list. Some who opted out, including journalist John Carreyrou, filed their own suits against Anthropic and other AI companies seeking up to $150,000 per work in statutory damages.¹¹

The New York Times v. OpenAI

The highest-profile individual copyright case is The New York Times Co. v. Microsoft Corp. and OpenAI, filed in December 2023 in the Southern District of New York. The Times seeks billions of dollars in damages and a permanent injunction that could require destruction of models trained on its content.¹²

In March 2025, Judge Sidney H. Stein partially granted and partially denied OpenAI’s motion to dismiss. Direct and contributory copyright infringement claims survived, along with certain DMCA claims. The court rejected the defendants’ statute-of-limitations defense for training data from 2019 to 2020. Common-law unfair competition claims were dismissed with prejudice.¹³

Discovery has been aggressive. In January 2026, the court affirmed an order requiring OpenAI to produce a de-identified sample of 20 million ChatGPT conversation logs. In April 2026, the judge granted the Times additional deposition time after finding OpenAI’s corporate designee unprepared. Summary judgment briefing wrapped up in April 2026, and a ruling is expected in the third quarter.¹² Twelve related copyright actions against OpenAI and Microsoft have been consolidated into a multidistrict litigation before the same judge, and the court has ordered production of an additional 88 million output logs in that proceeding.¹⁴

Music, Images, and Video: The Entertainment Industry Cases

Publishers were early, but the record labels, film studios, and visual artists are now catching up.

Andersen v. Stability AI, filed in 2023 in the Northern District of California, is the flagship case from visual artists against image generators. The artists allege that Stability AI, Midjourney, and DeviantArt infringed their copyrights by using their work to train models. Trial is scheduled for April 5, 2027.¹⁵ Getty Images v. Stability AI, filed in Delaware and separately in the United Kingdom, accuses Stability AI of copying millions of photographs.¹⁶

Disney, NBCUniversal, and Warner Bros. Discovery have moved against both Midjourney and Chinese AI company MiniMax, whose Hailuo AI service allegedly generates unauthorized images and video of copyrighted characters. The MiniMax complaint, filed in September 2025 in the Central District of California, alleges “willful and brazen” infringement and seeks up to $150,000 per infringed work. MiniMax filed motions to dismiss in April 2026.¹⁷¹⁸

On the music side, Sony Music, Universal Music Group, and Warner Records sued Suno and Udio in June 2024, coordinated by the RIAA. The labels alleged the AI music generators copied “decades worth of the world’s most popular sound recordings,” producing vocals “indistinguishable” from famous artists.¹⁹ Warner settled with both companies in late 2025 and formed licensing partnerships. UMG separately settled with Udio in October 2025, with plans for a joint AI music platform.²⁰ Sony’s litigation against both companies is still active. In spring 2026, UMG and Sony moved to add more than 61,000 sound recordings against Suno and more than 30,000 against Udio to the remaining cases.²¹

News publishers have followed a parallel track. Dow Jones and NYP Holdings, publishers of The Wall Street Journal and the New York Post, sued Perplexity AI in October 2024, alleging the search company copied articles into its retrieval index, produced verbatim reproductions, and generated fabricated content it falsely attributed to their publications. In August 2025, Judge Katherine Polk Failla denied Perplexity’s motion to dismiss in full, and the case is in discovery.²²

Beyond Copyright: Contract, Consumer, and Practice-of-Law Claims

Not every AI plaintiff is arguing copyright. Some are finding traction with older legal theories.

Reddit sued Anthropic in June 2025, alleging the company scraped user content from its platform without authorization to train Claude. Rather than claim copyright infringement, Reddit brought five California state-law claims: breach of contract, unjust enrichment, trespass to chattels, tortious interference with contract, and unfair competition.²³ Anthropic removed the case to federal court, arguing the claims were copyright infringement in disguise. On March 30, 2026, Judge Trina L. Thompson disagreed and sent the case back to state court, finding the claims “qualitatively different” from copyright. The breach-of-contract claim rested on Reddit’s user agreement restricting commercial scraping; the unfair competition claim alleged Anthropic publicly claimed to honor robots.txt while ignoring it during scraping.²⁴ Mediation was ordered by August 2026, with a jury trial set for February 2028.²⁵

A more unusual theory drives Nippon Life Insurance Company of America v. OpenAI, filed March 4, 2026, in the Northern District of Illinois. The case alleges ChatGPT effectively practiced law without a license. A former claimant uploaded attorney correspondence to the chatbot after a settled disability claim; the chatbot concluded she was being “gaslighted” by her lawyers; she fired them and tried to reopen the settled case, filing 21 motions, eight notices, and a subpoena, one of which included a fabricated court citation allegedly generated by ChatGPT.²⁶²⁷ Nippon Life alleges tortious interference with a contract, abuse of process, and unauthorized practice of law, seeking $300,000 in compensatory damages and $10 million in punitive damages. The complaint argues OpenAI marketed ChatGPT as capable of passing the bar exam without sufficient safeguards to prevent it from giving personalized legal advice.²⁸ OpenAI’s motion to dismiss is pending, with a status hearing scheduled for August 5, 2026.²⁹

Can AI Own What It Creates?

On March 2, 2026, the U.S. Supreme Court declined to hear Thaler v. Perlmutter, ending Stephen Thaler’s attempt to register purely AI-generated artwork for copyright protection. The lower courts had held that human authorship is a prerequisite for registration.¹⁴ The ruling does not settle what happens with mixed works. In the Copyright Office’s earlier Zarya of the Dawn determination, a graphic novel’s text and arrangement were held copyrightable, but its individual AI-generated images were not.¹⁶

AI Hiring Discrimination

A separate wave of litigation targets AI systems used to screen job applicants.

In Mobley v. Workday, Inc., plaintiffs allege that Workday’s AI applicant screening software discriminates on race, age, and disability. The Northern District of California ruled Workday can be liable as an agent of employers, finding “no legal distinction between delegating decisions to an ‘automated agent’ versus a ‘live human.'” In May 2025, Judge Rita F. Lin conditionally certified a collective action for applicants aged 40 or over who were rejected through the platform.³⁰ The potential class could number in the hundreds of millions given how widely the platform is used.³¹

In January 2026, a new class action, Kistler et al. v. Eightfold AI Inc., was filed against an AI hiring platform used by Microsoft, Morgan Stanley, Starbucks, and PayPal, among others. Two women with STEM backgrounds allege Eightfold scrapes personal data from social media and other online sources, then produces hidden “consumer reports” ranking candidates on their “likelihood of success” without applicants’ knowledge or consent. The complaint asserts violations of the Fair Credit Reporting Act, the California Investigative Consumer Reporting Agencies Act, and California’s unfair competition law.³² As of April 2026, the case is in the pleading stage.³³

In March 2025, the ACLU filed an administrative complaint with the EEOC and the Colorado Civil Rights Division against Intuit and HireVue on behalf of a deaf, Indigenous job applicant, alleging that HireVue’s AI interview tool discriminates against deaf applicants and those with non-standard speech patterns. Both companies denied the allegations, and HireVue stated that Intuit did not use an AI-backed assessment in the hiring process at issue.³⁴

Defamation, Hallucinations, and Harm to Minors

Courts have started taking on lawsuits over AI-generated falsehoods. In Walters v. OpenAI, a Georgia radio host sued after ChatGPT fabricated a legal summary falsely stating he had been charged with embezzlement. The court granted summary judgment to OpenAI, finding no proof of negligence or actual malice because the company had provided sufficient warnings about potential inaccuracies. In Battle v. Microsoft, an adjunct professor and CEO sued after Bing’s AI search summary conflated his identity with that of a convicted terrorist. A late 2024 ruling compelled the case to arbitration.³⁵

The cases against Character.AI have drawn the widest public attention. In October 2024, Megan Garcia sued in the Middle District of Florida after her 14-year-old son, Sewell Setzer III, died by suicide in February 2024. The complaint alleged the platform’s chatbot engaged the teenager in “abusive and sexual interactions,” encouraged harmful behaviors, and contributed to his mental health decline, asserting wrongful death, negligence, and product liability.³⁶ In January 2026, Character.AI, its founders Noam Shazeer and Daniel De Freitas, and Google (which had hired Shazeer) reached a mediated settlement with the Garcia family and four other families that had filed similar suits in New York, Colorado, and Texas. The terms were not publicly disclosed. Following the litigation, Character.AI prohibited users under 18 from back-and-forth conversations with chatbots and added safety features triggered by mentions of self-harm.³⁷

The CLEAR Act and Other Pending Legislation

Congress has started to move as well. In February 2026, Senators Adam Schiff and John Curtis introduced the Copyright Labeling and Ethical AI Reporting (CLEAR) Act. The bill would require any developer using copyrighted material in an AI training dataset to submit a detailed summary of every copyrighted work to the U.S. Copyright Office at least 30 days before commercial release. The Copyright Office would maintain a publicly searchable database. Copyright owners whose works were used without proper notice could sue in federal court, with penalties of $5,000 per violation, capped at $2.5 million per developer per year, plus injunctive relief and attorney’s fees.³⁸ The bill has been referred to committee and has not advanced further as of mid-2026.

Other Cases Working Through the Courts

Several additional cases are worth watching:

In Concord Music Group v. Anthropic, music publishers allege Claude reproduces copyrighted song lyrics; a preliminary injunction motion and a motion to dismiss are pending in the Northern District of California.¹⁵
In re Google Generative AI Copyright Litigation consolidates claims that Google scraped copyrighted works to train Gemini; a class certification hearing was scheduled for February 2026.¹
Doe v. GitHub, the Copilot litigation, alleges the AI coding assistant violates open-source licenses and the DMCA; the case is stayed pending an interlocutory appeal to the Ninth Circuit.¹⁵
Hendrix v. Apple is a consolidated class action targeting Apple’s OpenELM AI models, with leadership of class counsel pending as of early 2026.¹
Shareholder suits alleging “AI washing” have been filed against AppLovin Corporation, which allegedly attributed financial success to its AXON 2.0 AI ad platform rather than manipulative ad practices, and against Apple, which allegedly misled investors about the capabilities of its Apple Intelligence features.³⁰

What to Watch in Late 2026

Three decisions could reshape the field in the coming months. The Third Circuit’s ruling in Thomson Reuters v. ROSS would be the first appellate word on fair use in AI training. Judge Stein’s summary judgment ruling in the Times case could set the tone for the OpenAI multidistrict litigation and every case that follows it. And final approval of the Anthropic settlement would establish the going rate for pirated training data at a moment when other AI developers are watching closely to see whether their own datasets carry similar exposure.

1
Authors Alliance. AI Class Action Litigation Update: Books, Where Things Stand in Early 2026
2
Electronic Frontier Foundation. Two Courts Rule on Generative AI and Fair Use
3
Ropes Gray. A Tale of Three Cases: How Fair Use Is Playing Out in AI Copyright Lawsuits
4
Skadden. Fair Use and AI Training
5
Hollywood Reporter. Sarah Silverman Loses Key Issue in AI Lawsuit Against Meta
6
CourtListener. Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc
7
Bloomberg Law. Authors’ Copyright Class Action Certified Against Anthropic
8
Ars Technica. AI Industry Horrified to Face Largest Copyright Class Action Ever Certified
9
Authors Guild. What Authors Need to Know About the Anthropic Settlement
10
ClassAction.org. Bartz et al v. Anthropic PBC – Notice
11
Writer Beware. Anthropic Copyright Settlement April Update
12
AI Lawsuit Tracker. New York Times v. OpenAI
13
Justia. The New York Times Company v. Microsoft Corporation et al
14
Norton Rose Fulbright. AI in Litigation Series: An Update on AI Copyright Cases in 2026
15
BakerHostetler. Case Tracker: Artificial Intelligence Copyrights and Class Actions
16
New York State Bar Association. Copyright Law in the Age of AI
17
Variety. Disney, Warner Bros. Discovery and NBCUniversal Sue MiniMax
18
McKool Smith. AI Litigation Tracker
19
CNBC. Music Labels Sue AI Companies Suno, Udio for Copyright Infringement
20
The Vocal Market. Every AI Music Lawsuit Tracked
21
Music Business Worldwide. Suno Asks Court to Block UMG and Sony From Expanding Copyright Lawsuit
22
FindLaw. Dow Jones and Company Inc v. Perplexity AI Inc
23
Crowell & Moring. Reddit v. Anthropic Remanded to State Court
24
Loeb & Loeb. Reddit Inc v. Anthropic PBC
25
CourtListener. Reddit Inc v. Anthropic PBC
26
Georgetown Legal Ethics Journal. GPT Esquire: How the Nippon Case May Shape the Future of AI in Pro Se Litigation
27
American Bar Association. When Is a Settlement Not a Settlement: AI
28
Stanford Law School. Designed to Cross: Why Nippon Life v. OpenAI Is a Product Liability Case
29
CourtListener. Nippon Life Insurance Company of America v. OpenAI Foundation
30
Duane Morris. Class Action Review 2026
31
GovInfo. Mobley v. Workday Inc, Order on Motion for Collective Certification
32
Outten & Golden. Landmark Class Action Accuses Eightfold AI of Illegally Producing Hidden Credit Reports on Job Applicants
33
Akin Gump. AI Hiring Platform Faces FCRA Class Action Over Data Use
34
HR Dive. AI Intuit HireVue Deaf Indigenous Employee Discrimination ACLU
35
Columbia Law Review. Redefining Defamation: Establishing Proof of Fault for Libel and Slander in AI Hallucinations
36
NBC News. Character.AI Lawsuit Florida Teen Death
37
CNN. Character AI Google Settle Teen Suicide Lawsuit
38
IPWatchdog. CLEAR Act Would Establish Notice Requirements for Copyrighted Works in AI Training Data