A recent ruling by a US judge has significant implications for the intersection of artificial intelligence (AI) and copyright law. The decision addresses a lawsuit filed against the AI company Anthropic by three authors, including renowned mystery thriller author Andrea Bartz. The authors accused Anthropic of unlawfully using their literary works to train its Claude AI model, which has contributed to the company’s multi-billion dollar valuation.
Judge William Alsup presided over the case and concluded that Anthropic's utilization of the authors' books was "exceedingly transformative" and therefore permissible under US law. However, he denied Anthropic's attempt to dismiss the lawsuit, which means the company will face trial regarding its use of pirated copies to develop its extensive library.
Joining Bartz in the lawsuit were notable authors such as Charles Graeber, known for his non-fiction work The Good Nurse: A True Story of Medicine, Madness and Murder, and Kirk Wallace Johnson, the author of The Feather Thief. The judge's ruling indicates that Anthropic could be liable for damages amounting to up to $150,000 for each copyrighted work utilized without authorization. According to the judge, Anthropic has accumulated over seven million pirated books in its centralized library.
This ruling marks one of the first significant legal interpretations addressing how Large Language Models (LLMs) can ethically learn from existing literary materials. Judge Alsup noted that, similar to any aspiring writer, Anthropic's LLMs engage with works not to simply replicate them, but to create something distinct. "If this training process reasonably required making copies within the LLM or otherwise, those copies were engaged in a transformative use," Alsup stated.
Interestingly, the authors did not argue that the training led to infringing replicas of their works being produced for users of the Claude tool. Had they made such claims, the case could have taken a different direction altogether.
The ruling is part of a broader spectrum of legal battles concerning the AI industry’s use of various media, including journalistic articles, music, and video content. Recently, Disney and Universal have also initiated lawsuits against the AI image generator Midjourney, alleging piracy. Furthermore, the BBC is contemplating legal action regarding the unauthorized use of its content.
In light of these ongoing legal challenges, some AI companies have opted to negotiate licensing agreements with original creators and publishers to secure the rights to use their materials. Judge Alsup's ruling allows Anthropic to leverage a fair use defense, potentially influencing future legal judgments in similar cases.
In response to the ruling, Anthropic expressed satisfaction with the judge's acknowledgment of the transformative nature of its use of the works. However, the company disagreed with the decision to proceed to trial concerning the methods through which some of the books were acquired. Anthropic stated that it remains confident in its position and is currently evaluating its legal options moving forward.
The authors' legal representative opted not to provide any comments on the matter at this time, leaving questions about the potential outcomes of the upcoming trial unanswered.
