Authors push back as publishers and agents make claims on Anthropic settlement
- ID
- 21985
- Status
- summarized
- Published
- 07 Sep 2026, 4:47 AM
- Fetched
- 07 Sep 2026, 5:59 AM
- Provider
- TechCrunch
- Category
- technology
- Original URL
- https://techcrunch.com/2026/09/06/authors-push-back-as-publishers-and-agents-seek-share-of-anthropic-settlement/
- Source URL
- https://techcrunch.com/feed/
Summary
- Score
- 4.5
- Created
- 07 Sep 2026, 6:00 AM
- Tags
- Audience
- ai_ml_learnerssaas_founders
What happened
Authors in Anthropic's $1.5 billion copyright settlement are reporting that publishers are incorrectly claiming shares of their $3,000-per-title payments—either claiming rights to reverted works or taking 100% instead of the 50% they're entitled to for in-print titles. The settlement, finalized in July, covers nearly 500,000 titles and stems from a ruling that training AI on copyrighted material is fair use but pirating it is not.
Why it matters
The core legal precedent for builders is already set: training on copyrighted material is fair use, but acquiring that material through piracy is not. This article is mostly a payment-dispute follow-up between authors and publishers, so it doesn't require any change in how you build or ship. If you're sourcing training data, the actionable boundary remains: licensed or legitimately obtained content is fine; pirated content is not.
Discussion angle
The fair-use-vs-piracy distinction is now concrete with real dollar amounts ($3,000 per pirated title across 500,000 titles)—worth discussing what this means for data sourcing strategies if you're building AI products in Malaysia or Southeast Asia where copyright enforcement norms differ.