NYT accuses Microsoft of building supercomputer to enable AI copyright infringement

The New York Times has escalated its copyright lawsuit against Microsoft and OpenAI, now alleging that Microsoft's supercomputer was purpose-built to facilitate large-scale training on copyrighted works—with NYT journalism disproportionately included in the training data.

In an updated complaint, the NYT moved beyond treating Microsoft's infrastructure as generic cloud computing, instead claiming it was "unusually complex" infrastructure designed specifically to help infringe copyrights. The suit alleges that NYT articles were weighted more heavily during training because both companies sought to capture high-quality journalism that their models could then mimic in outputs.

The newspaper argues that this supercomputer served as the mechanism for "seizing copyrighted works without permission," and that Microsoft has unfairly profited from the arrangement. The complaint states: "Microsoft's deployment of Times-trained LLMs throughout its product line helped boost its market capitalization by a trillion dollars in the past year alone."

Evidence from user interactions#

The evidence presented includes ChatGPT sessions where users deliberately attempted to circumvent the NYT's paywall, with some requesting to see the "next paragraph" of articles and receiving substantial chunks of text in response. In other instances, the models "simply spit out several paragraphs" without prompting to do so.

The lawsuit, originally filed in 2023, has evolved to include allegations of market substitution and reputational harm, along with claims that ChatGPT outputs robbed Wirecutter writers of affiliate link commissions.