Judge approves $1.5B Anthropic settlement for pirated books used to train Claude
TL;DR Highlight
A judge approved Anthropic's $1.5B settlement over pirated books used to train Claude.
Who Should Read
Backend developer using LLM APIs who wonders about training-data copyright risk
Core Mechanics
- No article body was provided
- Per the title, a judge approved Anthropic's $1.5B settlement.
- It covers pirated books used to train Claude.
- No further details are in the source, so none are added.
Community Reactions
- Training is fair use, piracy is not
- Comments say Judge Alsup's earlier ruling held this.
- One commenter said the issue was piracy, not training itself.
- About $3k per eligible title
- A comment says payout per book is $3k.
- Traditional contracts reportedly split it between author and publisher.
- Class counsel fees were halved
- A comment says fees dropped from 12.5% ($187.5M) to 6.8% ($101M).
- Three class representatives reportedly get $15k each.
- Many called the sum too small
- Several commenters called it a cost of doing business.
- Some argued for royalties instead of a one-time payment.
- Installments tie payers to Anthropic's survival, one opinion says
- A commenter says the $1.5B is paid in installments with fees paid alongside.
How to Apply
- List the sources of your RAG/training data
- Within 30 minutes, tabulate the source and license of each document used for fine-tuning or RAG.
- Flag data with unknown provenance.
Terminology
Related Papers
The Judge Can Debias Itself: Awareness and Self-Debiasing for Reliable LLM-as-a-Judge
Warning an LLM judge about bias or letting it self-critique improved 20 of 21 model–bias combinations.
90% of US working hours are spent on tasks not covered well by AI benchmarks
Mapping 29 benchmarks to job tasks shows 90% of US working hours lack good coverage
Postgres 19 delayed after AI reviewers find more bugs than humans can fix
LLM code review surfaced so many bugs that Postgres 19 slipped to November and 8-10 features were reverted
Show HN: Gemini 4 Argon plays StarCraft
LLMs wrote StarCraft bots in one hour; GPT-6 Astra scored highest.
A Plausible Explanation Is Enough to Mislead Multimodal Judges
Adding plausible explanations to fixed answers cut MLLM judges' wrong-answer rejection by 12.7–13.8 points.
Neural networks are clean
Looking at neurons directly shows most have one main job, challenging superposition.