A US court has ruled in favor of Anthropic, a prominent AI firm, amidst a lawsuit filed by three authors, including best-selling mystery thriller writer Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson. The lawsuit, filed last fall, alleged that Anthropic had infringed on their works by using them to train its Claude AI model. The presiding judge in the case, the honorable William Alsup, made headlines himself. He issued a strongly-worded mixed verdict that could make the biggest change to copyright law in the area of artificial intelligence.
In his ruling, Judge Alsup decided that Anthropic’s use of the authors’ books was “super duper transformative.” This designation strengthened the company’s hands in making a “fair use” defense on the basis of US law. It permits for the use of copyrighted material without permission, so long as those uses are ultimately deemed transformative. Despite this victory, the judge acknowledged that Anthropic had violated the authors’ rights by saving copies of their books without authorization.
Anthropic, for its part, is said to host more than seven million illegally reproduced texts in its “central library.” This part of the case introduced some very important issues surrounding the legality of using copyrighted works to improve existing generative AI training. Judge Alsup soundly rejected Anthropic’s request to dismiss the case. In spite of that, he allowed the fair use defense to proceed, paving the way for possible future legal precedents in this still-maturing field of law.
The court’s ruling highlights the need for a clearer understanding of the legal implications at the intersection of artificial intelligence and IP rights. Judge Alsup noted the intention behind Anthropic’s training process, stating, “Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works, not to race ahead and replicate or supplant them — but to turn a hard corner and create something different.”
Moreover, he added, “If this training process reasonably required making copies within the LLM or otherwise, those copies were engaged in a transformative use.” Future litigants should take heart that these statements will guide how courts read similar cases going forward.
Anthropic was disappointed to hear of the court’s decision. They seem unhappy to be forced to allow the trial to go forward concerning how those books were obtained and deployed. The company is facing down a judge’s ordered trial. It needs to answer a myriad of questions and justify its practices with regard to the development of the Claude AI model.