Last weekend, Anthropic announced that it had settled a class action lawsuit with authors whose works were pirated to train artificial intelligence (AI) models. The $1.5 billion settlement will cover more than 500,000 works that were found to have been pirated without the permission of the intellectual property owner.
However, California judge William Alsup rejected the settlement, saying that the affected authors would now only have to pay $3,000 for each pirated book. This was seen as unfair, as the full list of works and authors affected by piracy has not been provided.
Alsup also asked that the lawyers notify the class action members of the settlement and produce a form to accept the compensation or not. The authors must also agree not to sue Anthropic in the future.
Alsup previously found that Anthropic was not guilty of infringing on authors' intellectual property when using their work to train AI because it was considered fair use. Anthropic was only guilty of storing 7,000,000 pirated books in a central library without all of them being used to train AI models.
