TLDRs;
- Judge William Alsup has paused Anthropic’s $1.5B book piracy settlement, citing transparency and fairness concerns.
- The proposed deal would have paid authors $3,000 per pirated book, with nearly 465,000 titles affected.
- Unlike fair use disputes, this case focuses on alleged illegal downloads from piracy sites like Library Genesis.
- The ruling could force AI firms to adopt clearer licensing practices and offer authors stronger protections in future deals.
A federal judge in the United States has put the brakes on Anthropic’s proposed $1.5 billion settlement over allegations that its artificial intelligence models were trained on pirated books. T
Judge William Alsup, presiding over the case, expressed strong reservations about the settlement’s structure, particularly its lack of transparency and the role of class action attorneys in crafting the deal. He emphasized the need for greater author participation and insisted on more details regarding the claims process and the scope of works involved.
The lawsuit, filed by a coalition of U.S. authors, accused Anthropic of training its large language models on content scraped from illegal sources such as Library Genesis. An estimated 465,000 titles were said to be affected, with the proposed settlement offering $3,000 per book in compensation to authors and publishers.
Judge Rejects “Behind Closed Doors” Deals
Judge Alsup raised concerns that the agreement might have been created “behind closed doors” by lawyers eager to reach a quick resolution rather than ensuring fair compensation for authors.
“This court will not allow a settlement to be forced down the throats of authors without their full knowledge and consent,” Alsup noted during hearings.
He also requested a “drop-dead list” of the allegedly pirated titles to clarify exactly which works would fall under the deal.
Such scrutiny is unusual in class action cases, where judges often approve broad settlements designed to end litigation efficiently. Alsup’s stance suggests that copyright disputes involving AI may require more rigorous oversight to safeguard creator rights.
Why the Case Stands Out
The Anthropic dispute differs from other AI copyright lawsuits because it centers not on the fair use of purchased books but on claims of outright piracy.
While debates continue about whether AI training on licensed or purchased works qualifies as fair use, this case focuses on content sourced from unauthorized sites.
Legal experts note that the $1.5 billion settlement would have resolved Anthropic’s “past liability related to piracy” without reshaping the broader legal framework for AI training. By halting the deal, the court has left open the question of how AI companies can compensate creators while protecting themselves from future lawsuits.
Implications for the AI Industry
The decision signals that U.S. courts are increasingly sensitive to the interests of human creators as AI development accelerates. If upheld, Alsup’s demands could push companies like Anthropic, OpenAI, and others to adopt clearer licensing agreements and greater transparency in data sourcing.
For authors, the ruling may provide an opportunity to negotiate stronger terms and secure better recognition of their work in an AI-driven economy. However, it also prolongs uncertainty, as a final resolution remains distant.
