Judge skewers $1.5B Anthropic settlement with authors in pirated books case (apnews.com)

A federal judge has expressed serious reservations about the $1.5 billion settlement between AI company Anthropic and authors who claimed their copyrighted books—nearly 465,000 in total—were illegally pirated to train Anthropic’s Claude chatbot. Judge William Alsup criticized the settlement for its potential pitfalls, particularly around ensuring all affected authors are properly notified and compensated. He has set deadlines for clarifying the exact number of pirated works and refining the claims process, with a follow-up hearing scheduled for late September to determine whether the settlement can proceed or if the case will move to trial. This case holds significant implications for the AI/ML community as it highlights ongoing legal challenges surrounding the use of copyrighted materials in AI training data. While the judge previously ruled that training AI on copyrighted books isn’t inherently illegal, the method by which Anthropic acquired the data—via pirate websites—raises serious legal and ethical concerns. The outcome could shape how future AI models approach data sourcing and establish financial liabilities for companies using unauthorized content, reinforcing the importance of transparency and rigorous data auditing in AI development. Judge Alsup also expressed concern over potential undue influence by groups like the Authors Guild and the Association of American Publishers, cautioning that authors might be pressured into settlement agreements without full understanding. The case underscores a broader tension between protecting intellectual property rights and fueling AI innovation, potentially setting a precedent for how AI training datasets are curated and compensated moving forward.