AI Risk Wire

Cases › copyright or IP violation › LR-2026-0346

NZ authors receive payout after AI chatbot trained on pirated books

Preliminary Why: Publisher not yet named · subject to change

Status: no action reported

Severity moderate
Reported 3 Dec 2025
Developer and model Anthropic

No source states: loss or records, agent tool, could do, deployer, industry.

Severity moderate: Test: moderate, because data of others was touched and the scope is bounded: only books registered with the US copyright office were eligible for the settlement

Anthropic agreed to pay up to $1.5 billion to settle a class action in US courts over claims that it used millions of pirated books to train its large language models. Commentator Peter Griffin said the settlement pivoted on the company's use of two book piracy websites to source content, and that the ruling did not counter the use of intellectual property for training AI models without permission. Only books registered with the US copyright office were eligible for the settlement, so authors of many titles will not be compensated. Author Catherine Chidgey said she received an email saying her books had been identified as caught up in the case, and that authors were offered US$3000 for each title accessed. The New Zealand Society of Authors said about 1.5 million of about 7.5 million pirated books would be eligible for compensation.

What happened

1 · What it was doingThe company trained its large language models on millions of books that were allegedly pirated.
2 · What went wrongThe works were accessed from two websites that held pirated copies, which the author says were not legally accessed.
3 · What it reached or didThe pirated library was reported to hold about 7.5 million books, with about 1.5 million eligible for the settlement.

What was done

No source states an action or fix.

Risks

Copyright and IP
Primary7 incidents

Read from the sources; not yet reviewed by a person.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Model

Copyright safeguards · Marking machine-made output

Application

Output filtering and redaction

Similar incidents

Judge approves Anthropic's $1.5 billion copyright settlement with authors2026-07-20
Elvis Presley rights holders sue Anthropic over 274 songs used in training and reproduced by Claude2026-10-25
Claude training-data study: blackmail and code sabotage in altered research models2026-10-31
AI-assisted exploitation of PHP2Shell and Fragnesia flaws to breach 25 servers across four countries2026-10-29

All cases of this kind ›

Evidence

No source states: how it was caught, when it happened, when it was found, how many records were exposed, a count of the other parties affected, any court case, the cause, what set it off, what was done.

Show the full account: 4 items from the primary sources

What happened, in order

  1. 2027 The current review of the Copyright Act was expected not to address AI issues until its second stage.

Not dated in the sources:

  • Anthropic agreed to pay up to $1.5 billion to settle the class action.
  • Author Catherine Chidgey was emailed that her books had been identified as caught up in the case.
  • The society encouraged members to register their books with the US copyright office for future claims.
Show every field, with its source and grade

The AI system

Developer
Anthropic
C one secondary source states it

What happened

Earliest source we hold
2025-12-03
D inferred by us from the described behaviour, reason shown
What went wrong
copyright or IP violation
C one secondary source states it
primary risk
RK17
D inferred by us from the described behaviour, reason shown

Impact

Severity
moderate
D inferred by us from the described behaviour, reason shown
Kind of harm
intellectual property
C one secondary source states it
Why this severity
Test: moderate, because data of others was touched and the scope is bounded: only books registered with the US copyright office were eligible for the settlement
D inferred by us from the described behaviour, reason shown

Context

Country
United States
C one secondary source states it

Notes on gaps

The number of affected authors is not stated; the 7.5 million and 1.5 million figures are book counts. The settlement amount and per-title payment are press-reported.

LR-2026-0346 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON