The Thriller Writer Who Took on a Tech Giant (www.nytimes.com)

Author Andrea Bartz was one of three named plaintiffs in a class-action suit that culminated in a record $1.5 billion settlement with Anthropic after a judge found the company had illegally downloaded and stored hundreds of thousands—by some accounts millions—of pirated books used to train its chatbots. Bartz, who discovered her own novels listed in a searchable database and felt a bot-generated piece in her voice, joined other writers to pursue damages; the settlement offers $3,000 per affected work (with publishers taking up to half depending on contracts) and potential service awards of up to $50,000 for named plaintiffs. The parties agreed to settle after the court concluded Anthropic knowingly acquired pirated texts. The case is significant because it’s the largest copyright payout touching AI training data and signals real financial risk for companies that ingest pirated corpora. But its legal impact is mixed: the judge also held that using non‑stolen books for model training can be fair use as a “transformative” activity, a conclusion authors and publishers contest. Because the matter settled rather than producing a definitive trial precedent, the ruling doesn’t resolve broader questions—such as how to value training data, how royalties should be apportioned, or how courts will treat AI‑generated works—but the settlement could pressure firms to license datasets and sharpen compliance around data provenance.