There are several ongoing cases against Anthropic right now, so let’s be a little more specific. When people talk about “the Anthropic case,” they usually mean Bartz v. Anthropic, where authors sued Anthropic for taking their copyrighted material without consent and using it to train their large language model (LLM), Claude. Recent developments have brought this case back into the news.
First, let’s do a brief recap for those who haven’t been following the case. This is a class-action lawsuit, meaning a small group of authors filed the lawsuit on behalf of all authors harmed by Anthropic’s actions. Anthropic attempted to avoid a full trial by asking the judge to make a summary judgment, i.e., to issue a ruling based on the facts of the case, without a full trial. In June of 2025, the judge issued a partial summary judgment. The judge said that using copyright material to train an LLM could be considered fair use. This caused quite a stir, because this would be a major win for AI companies if a higher court agreed with the judge’s reasoning. (A district court ruling can only affect the parties to the case; an appellate court ruling is binding on all the courts below it.) However, the judge said that how Anthropic acquired the copyrighted content mattered. Because much of the copyrighted content came from pirated websites, the acquisition method could lead to a finding of copyright infringement. The court said this aspect of the case had to go to trial.
In the end, Anthropic decided it still didn’t want to go to trial over this. They worked out a settlement agreement with the parties, gave the rest of the class an opportunity to join and/or object to the settlement, and offered it to the court. In July 2026, the court approved the settlement, with a few modifications that benefit the authors. Those authors who joined the settlement should start receiving money in the not-too-distant future.
But what does this mean from a legal perspective? Unfortunately, not a whole lot. Because the parties settled, there is no official ruling, and because it never went to an appellate court, the possible fair use finding is not binding on any other court. The question of whether an AI company can use copyrighted content to train their LLM without permission still waits to be decided.
But don’t despair. Several other cases address the same issue, and there are good odds that at least one will reach an appellate court, if not the Supreme Court.
If you have questions about this Anthropic case, or about AI and copyright generally, please feel free to reach out to me at kaway@kawaylaw.com.
