Screw the money — Anthropic’s $1.5B copyright settlement sucks for writers (techcrunch.com)

Anthropic has agreed to a historic $1.5 billion settlement in a class-action lawsuit brought by around half a million writers over its unauthorized use of copyrighted books to train its AI language model, Claude. While this marks the largest payout in U.S. copyright history, many see it as a minor penalty for Anthropic, which recently raised $13 billion and continues to benefit from training on vast amounts of copyrighted material scraped from the internet and “shadow libraries.” The case underscores ongoing tensions between AI developers eager to feed their models diverse text data and creators seeking compensation for the use of their work. The lawsuit, Bartz v. Anthropic, hinged not on whether AI training on copyrighted texts is legal—federal judge William Alsup ruled in favor of Anthropic, citing the "transformative" nature of such use under the century-old fair use doctrine—but on the company’s illicit methods of obtaining data. This ruling sets a significant legal precedent amid numerous similar cases against major tech players like Meta, Google, and OpenAI, shaping how courts may approach intellectual property in AI training. Despite the settlement averting a trial, the outcome raises critical questions about fair compensation, creative ownership, and the evolution of copyright law in the age of large language models.