AI copyright news, graded by source

At the Bartz final approval hearing, class counsel describes the settlement's non-money terms: pirated datasets destroyed, past conduct only, no release of output claims

At the May 14, 2026 hearing, class counsel told the court that the claims rate was 92.77%, or 447,576 works, and that there were 350 valid opt-outs covering 1,802 works. He described the settlement as including destruction of the pirated datasets and a certification by Anthropic that material from the pirate sites was not used in commercially released models, and said the release covers "past conduct only and not including output claims." He said the owner of a work that is not on the works list is not in the class and has released nothing, and that 26 of 53 objections came from people asking to be added to the list. Anthropic's lawyer, asked if she wished to add anything, said she did not. Objectors told the court the list missed works registered in groups or under a pen name. The judge took the matter under submission and granted final approval on July 20, 2026. These are statements at a hearing, not the settlement agreement's text.