AI and copyright: OpenAI and Microsoft face the decisive test of market harm
This battle of the experts goes far beyond a procedural dispute. It should shed light on how the authors and publishers intend to demonstrate that OpenAI’s models do not merely learn from their works, but produce content capable of acting as an economic substitute for them. This is precisely the ground on which the authors in the Kadrey v. Meta case failed in June 2025. Judge Vince Chhabria had recognised the highly… transformative nature of Llama’s training and ultimately granted Meta the benefit of fair use . However, this victory rested less on a general endorsement of unauthorised training than on the inadequacy of the evidence produced by the claimants. The plaintiffs had primarily argued that Llama could reproduce extracts from their books and that the free use of their…