AI copyright lawsuits tracker

Cases at a glance

Bartz v. Anthropic

Authors alleged Anthropic trained Claude on pirated books. Judge William Alsup ruled that training on legally purchased books could be fair use, but that Anthropic was not entitled to use pirated copies to build a central library. The case then settled for $1.5 billion, about $3,000 per book, the largest copyright settlement on record. Judge Araceli Martínez-Olguín gave final approval in July 2026 (JURIST).

What it means: the source of data matters as much as its use. Lawfully acquired, licensed data is the low-risk route.

Kadrey v. Meta

On June 25, 2025, Judge Vince Chhabria granted Meta partial summary judgment, finding that training Llama on the plaintiffs' books was fair use because the plaintiffs did not offer meaningful evidence of market harm. He stressed the ruling did not mean all training on copyrighted works is lawful, and suggested plaintiffs with better evidence of market dilution could win (Akin Gump).

What it means: evidence of a licensing market strengthens rights holders' arguments, which is one reason licensing deals are being publicized.

Thomson Reuters v. Ross Intelligence

Ross built an AI legal search tool using material derived from Westlaw. The district court rejected Ross's fair use defense. In September 2026 the Third Circuit upheld that ruling in what was reported as the first appellate decision in the wave of AI training cases; its reasoning was filed under seal (Talking Biz News).

The New York Times v. OpenAI and Microsoft

The Times sued in late 2023, alleging millions of its articles were used to train ChatGPT and other tools. In June 2026 it amended its complaint, modifying one claim against Microsoft and dropping another against OpenAI (GigaLaw). The case is ongoing.

Getty Images v. Stability AI (UK)

Getty dropped its primary infringement claims during trial. On November 4, 2025, the High Court dismissed its secondary copyright infringement claims and found some trademark infringement, which the judge called extremely limited in scope (Cleary Gottlieb).

GEMA v. OpenAI (Germany)

The Munich Regional Court found that song lyrics were embedded in GPT-4 and GPT-4o and could be reproduced with simple prompts. It accepted that training generally falls within Germany's text and data mining exception, but held that memorizing the lyrics went beyond it. It granted injunctive relief, disclosure and damages. OpenAI announced an appeal (Bird & Bird-on-copyright-and-ai-training)).

Reddit's cases

Reddit sued Anthropic in San Francisco Superior Court on June 4, 2025, for breach of contract, unjust enrichment, trespass to chattels and unfair competition. A federal judge sent the case back to state court on March 30, 2026 (Crowell & Moring). In October 2025 Reddit sued SerpApi, Oxylabs, AWMProxy and Perplexity in New York, alleging they scraped Reddit content through Google search results (Claims Journal).

What it means: contract and platform terms can restrict data use even where copyright might not. Check the terms that govern any data you license.

Disney, Universal and Warner Bros. v. Midjourney

The studios began suing Midjourney in June 2025 and consolidated their cases in November 2025. In May 2026 the parties were in a discovery dispute before a magistrate judge (Graphic Policy).

How this affects data deals

• Buyers want warranties that data was lawfully obtained and that the seller holds the rights granted.
• Sellers with clean rights and documentation have leverage.
• Licenses should state what happens if a third party claims rights in the data.

Read AI data license agreement and fair use.