Authors Guild asks court to rule OpenAI's training was piracy
The Authors Guild and 13 named authors filed for summary judgment arguing OpenAI trained ChatGPT on torrented Library Genesis books and can't claim fair use.
Developing · day 5 · first reported Thu, Sep 10 2026
The Authors Guild just asked a federal court to rule, as a matter of law, that OpenAI's training was piracy, not fair use. Thirteen named authors -- including John Grisham, George R.R. Martin and Jodi Picoult -- back the motion, covering 194 titles and seeking a liability finding only, not damages yet. TorrentFreak's read of the unredacted filing is sharpest: OpenAI allegedly renamed its 'Libgen1'/'Libgen2' training sets to 'Books1'/'Books2' in GPT-3 documentation, then deleted the files in summer 2022. This ruling will set the fair-use bar every other AI-training case leans on.
John Grisham, George R.R. Martin, Jodi Picoult and ten other named authors have joined the Authors Guild in asking a New York federal court to do something no ruling has done yet: find, as a matter of law, that OpenAI's training was piracy -- not the fair use the company has argued in courtrooms for two years.
The motion covers 194 titles and asks for a liability finding only. No damages figure is on the table yet -- that fight comes later. Opposition briefs from OpenAI and Microsoft are due early October, with reply briefs from the authors' side following in early November.
The strongest material in this week's coverage comes from TorrentFreak, working from the less-redacted portions of the filing. It quotes the motion directly: OpenAI 'did not even buy the books it used,' instead torrenting them from Library Genesis. The unredacted sections allege the company then renamed its own training datasets -- internally labeled 'Libgen1' and 'Libgen2' -- to the blander 'Books1' and 'Books2' in its GPT-3 documentation. By summer 2022, as legal scrutiny mounted, OpenAI deleted the LibGen files outright -- the only training corpus, per the authors' filing, that it has ever removed.
Publishers Weekly's account, working from the fuller but more heavily redacted version, doesn't carry that renaming-and-deletion detail. What it does add is the complete plaintiff roster -- Baldacci, Branch, Connelly, Day, Franzen, Golden, Greer, Grisham, Hwang, Martin, Picoult, Schiff and Shapiro -- plus a clearer look at OpenAI and Microsoft's counter-argument, which leans on the Google Books and Kadrey v. Meta precedents to argue training remains transformative fair use. Publishers Marketplace adds the useful context that this motion is running in parallel, before the same judge, with a separate summary-judgment fight brought by news plaintiffs led by the New York Times.
There's no form to fill out and no deadline for you this week. But don't mistake 'nothing to do' for 'nothing at stake.' If the court finds OpenAI's use wasn't fair use, every pending and future author lawsuit against an AI company gets more leverage overnight. If it goes the other way, it likely shuts that argument down for the current wave of suits. Worth bookmarking November, when the reply briefs close and this starts moving toward an actual ruling.
Why it matters to you: If the court finds OpenAI's use wasn't fair use, it strengthens every other author's leverage in pending and future AI-training disputes; if it doesn't, it likely closes that door for the current wave of suits.
This story ran in the Coral Wire edition of Wed, Sep 23 2026, when the Coral Index stood at 35/100 (Headwinds).
Coverage — 3 outlets
Every outlet found reporting this story.
- Authors Guild, Co-Plaintiffs Seek Summary Judgment in OpenAI Casepublishersweekly.com
- Authors, OpenAI File for Summary Judgement In New York Infringement Caselunch.publishersmarketplace.com
- OpenAI's ChatGPT Was Built on Concealed 'Mass Piracy', Authors Tell Courttorrentfreak.com
Earlier in this story
- Tue, Sep 22 2026OpenAI executives admitted knowingly copying your work illegally
- Sun, Sep 20 2026OpenAI, Microsoft knew AI training harmed web content
- Sat, Sep 19 2026OpenAI, Microsoft knew AI training harmed web content
- Thu, Sep 10 2026Authors Guild seeks summary judgment against OpenAI copyright case
Also in this edition
- Authors balance AI tools while protecting authentic voicePlatform — Practical guidance helps you leverage AI without compromising originality or reader trust.
- Anthropic settlement payouts cut to $2,204 as claims disputes mountLegal — Authors' first Anthropic settlement payments, due November 1-15, work out to $2,203.56 per title after publishers and literary agents were found claiming shares they aren't entitled to.
- Court files prove AI firms knowingly copied your workLegal — Unsealed documents strengthen authors' legal position in ongoing copyright infringement cases.
- OpenAI executives admitted copying authors' work illegallyLegal — Direct evidence of intentional infringement could shift compensation negotiations in your favor.
- Indie authors' sales grew 64% on one platform as publishers grew 24%Market — PublishDrive's own platform data shows independent authors' sales value growing 64% year-over-year versus 24% for publishers, with author share of romance-fantasy revenue rising sharply.
- AI romance author Coral Hart targets 2026 marketWire — Direct competitor entry signals growing viability of AI-assisted romance publishing.
- New AI models, price wars reshape author tool costsAI — Claude Opus, GPT-6 updates affect your editing and writing software economics.
- EU's AI disclosure law is now live for AI-assisted booksLegal — Since August 2, 2026, the EU AI Act's Article 50 requires labeling AI-generated or AI-narrated content reaching EU readers, with fines up to EUR 15m or 3% of global turnover for non-compliance.