Authors, newspapers and reference companies say AI developers copied their work without permission to train models. The developers argue that training is fair use. That argument is now being tested in several US courts at once, and the outcomes will decide whether AI companies pay for the books and journalism they learn from, and how much.
The signals so far point both ways. In Kadrey v. Meta, the judge found the authors had failed to show market harm, Bloomberg Law reported. In September the Third Circuit ruled in Thomson Reuters v. ROSS Intelligence that copying Westlaw headnotes to train a non-generative legal search tool was not fair use. The first authors’ class action against Anthropic ended in a $1.5 billion settlement, according to Bloomberg Law, and Korean publishing houses fear the per-work figure could become a price benchmark, Seoul Economic Daily reported.
The pressure is now on Meta and OpenAI. About 300 authors have filed individual claims against Meta, and their lawyers plan claims for close to 1,000 more, Bloomberg Law reported. A magistrate judge has ordered Meta to name everyone who approved downloading “shadow library” datasets, a discovery step that makes no finding of liability. In New York, a newly unsealed filing in the consolidated case against OpenAI alleges Microsoft told staff to keep a crawler project secret, and USA Today Co. has sued seeking more than $250 million. These remain allegations.
Alongside the litigation runs a market for licences. CNN has signed a multi-year AI deal with Google, a coalition of news organisations is pressing for usage rights, and Meta dropped a line from its business terms that would have allowed crawled content to train its products.
Where things stand: no appeals court has yet ruled on fair use for generative AI training, and the US Justice Department has told the OpenAI court that training, in its view, is lawful. That is a statement of interest, not a ruling.
Timeline
October 2026
- 22 Oct 2026: Anthropic Seeks Fair-Use Ruling on Claude Lyrics as Judge Weighs First AI TrialAnthropic asked a US judge to rule that training Claude on song lyrics is fair use. The judge did not rule, but weighed whether…
- 20 Oct 2026: Judge Orders Meta to Name Who Approved Its Shadow Library DownloadsA US judge has ordered Meta to name everyone who approved obtaining shadow libraries used for AI training. It is a discovery order, not…
- 20 Oct 2026: Around 300 Authors Bring Individual AI Copyright Claims Against MetaAbout 300 authors have sued Meta, alleging it trained AI on pirated copies of their books. They are pursuing individual claims, not a class…
- 18 Oct 2026: Google Is Negotiating to Buy Korean Books for AI Training, Seoul Paper ReportsGoogle has been negotiating with major Korean publishers to buy books for AI training, Seoul Economic Daily reports. Google Korea did not respond to…
- 17 Oct 2026: Publishers’ Filing Says Microsoft Told Staff to Keep ‘Project Mango’ SecretA newly unsealed filing by news publishers alleges Microsoft told staff to keep Project Mango, a crawler OpenAI paid it to run, secret. No…
- 16 Oct 2026: Meta Drops ‘Train’ Line From Its Terms as Publishers Press EU and UK RegulatorsMeta removed the 'train' line from its new Commercial Terms. European publishers now want Brussels and the UK's CMA to stop the terms before…
- 15 Oct 2026: CNN Signs a Multi-Year AI Partnership With GoogleCNN announced a multi-year AI partnership with Google for experimental features and content rights. No financial terms. Reuters and Variety add context.
- 13 Oct 2026: The New York Times, The Economist and AFP Join More Than 40 Publishers in SPUR, a Coalition on AI Usage RightsThe New York Times, The Economist and AFP join more than 40 publishers signing up to SPUR, a coalition setting standards for AI use…
- 10 Oct 2026: USA Today and 13 Sister Companies Sue OpenAI: What the Complaint AllegesOn 8 October 2026, USA Today Co. and 13 affiliated companies, covering 19 publications, filed a copyright lawsuit against seven OpenAI entities in federal…
- 5 Oct 2026: 24 Articles from Beijing: How China’s Top Court Wants AI Disputes HeardOn 7 September 2026 China's Supreme People's Court published Opinions on hearing AI-related disputes: 24 articles covering fault-based liability, notice-and-takedown, training-data evidence, open-source safe…
- 5 Oct 2026: Not Fair Use: What the Third Circuit Just Said About Training AI on Westlaw HeadnotesThe Third Circuit affirmed that 2,243 Westlaw headnotes are copyrightable and that ROSS's use of them to train a non-generative legal search tool was…
Key moments
- 19 Oct 2026: A magistrate judge orders Meta to name who approved its shadow library downloads
- 19 Oct 2026: About 300 authors file individual copyright claims against Meta, Bloomberg Law reports
- 18 Oct 2026: Google has been negotiating to buy Korean books for AI training, Seoul Economic Daily reports
- 17 Oct 2026: Meta removes the “train” sentence from its new Commercial Terms after a backlash, The Verge reports
- 16 Oct 2026: An unsealed filing alleges Microsoft told staff to keep “Project Mango” secret
- 15 Oct 2026: CNN signs a multi-year AI partnership with Google; no financial terms disclosed
- 12 Oct 2026: The New York Times, The Economist and AFP join the SPUR AI rights coalition, Reuters reports
- 8 Oct 2026: USA Today Co. and 13 affiliates sue OpenAI in New York, seeking more than $250m
- 29 Sep 2026: The Third Circuit rules that ROSS’s use of Westlaw headnotes was not fair use
- 7 Sep 2026: China’s Supreme People’s Court issues 24-article guidance on AI disputes, leaving copyright in AI output open
- 1 Sep 2026: The Justice Department tells the OpenAI court that training “does not violate copyright laws”
- 20 Jul 2026: The Bartz v. Anthropic class settlement receives final approval
