Nadella Calls Out AI Distillation Hypocrisy as Anthropic Pays $1.5 Billion in Copyright Case
Updated
Updated · Computerworld · Jul 28
Nadella Calls Out AI Distillation Hypocrisy as Anthropic Pays $1.5 Billion in Copyright Case
3 articles · Updated · Computerworld · Jul 28
Summary
Satya Nadella said major AI companies are hypocritical for defending “fair use” in training models on public data while imposing restrictive terms on distillation by smaller rivals.
Distillation uses one model’s outputs to train another, a practice analysts say is widespread across AI and one Elon Musk has acknowledged xAI used against OpenAI.
The dispute has sharpened as Chinese AI firms use distillation to narrow the gap with U.S. leaders, prompting OpenAI and Anthropic to press for coordinated legal and policy action.
Anthropic’s $1.5 billion settlement with authors last week underscored the copyright pressure on the industry, even as Microsoft and other AI companies still face multiple lawsuits from publishers and writers.
Anthropic paid $1.5 billion for pirating books, but if AI legally buys your book to replace you, can you even stop it?
With courts allowing AI to train on lawfully purchased texts, is the concept of intellectual property fundamentally broken in the generative era?
Landmark $1.5 Billion Bartz v. Anthropic Settlement: Reshaping AI Training Data, Copyright Law, and Global Industry Standards
Overview
The Bartz v. Anthropic case began when Anthropic used millions of pirated books from shadow libraries to train its Claude AI models, leading bestselling authors to file a class-action lawsuit. Judge Alsup ruled that training on legally acquired books was fair use, but using pirated copies was not, exposing Anthropic to massive damages. To avoid a risky trial, Anthropic settled for $1.5 billion and agreed to destroy all pirated data, but strict eligibility rules meant most independent and international authors received nothing, while major publishers gained large payouts. This settlement pushed the AI industry toward expensive licensing deals, but legal uncertainty remains as courts and global regulations diverge sharply.