Wed, 25 Jun 2025

Wed, 25 Jun 2025 Judge backs AI firm over use of copyrighted books

A US court has ruled Anthropic was not breaching copyright rules when it trained its AI model on books.

* Judge William Alsup calls Anthropic's use "exceedingly transformative", allowing it under US law
* Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson accused firm of stealing their work to train Claude AI model
* Firm could face up to $150,000 in damages per copyrighted work
* Judge rejects request to dismiss case, paving way for trial over use of pirated copies
* Anthropic backed by Amazon and Alphabet, holds more than 7 million pirated books in "central library"
* Decision is among the first to weigh in on question of how Large Language Models can learn from existing material without violating copyright law.
  >>