Updated
Updated · The New York Times · Aug 24
Judge Finds Anthropic Infringed Copyright by Downloading Millions of Books for AI Training
Updated
Updated · The New York Times · Aug 24

Judge Finds Anthropic Infringed Copyright by Downloading Millions of Books for AI Training

3 articles · Updated · The New York Times · Aug 24

Summary

  • A judge ruled Anthropic violated copyright law by illegally downloading and storing millions of copyrighted books used to train its Claude chatbot.
  • The court drew a line between two data-gathering methods, finding training on physical books Anthropic bought was fair use while the pirated digital library was infringement.
  • The case grew out of a class action by authors including Kirk Wallace Johnson and Andrea Bartz, who argued Anthropic used their books without permission or payment to build a commercial writing tool.
  • The ruling adds to mounting legal pressure on AI companies over training data, even as courts still differ on whether some forms of model training qualify as fair use.

Insights

With a historic settlement punishing AI piracy, will the skyrocketing cost of original human data ultimately bankrupt the generative AI boom?
If AI chatbots are quietly degrading by recycling synthetic text, how soon will the tools we rely on become completely useless?
As models face catastrophic collapse from cannibalizing their own outputs, could human writers become the most valuable commodity in tech?