Unsealed court filings show Microsoft and OpenAI staff warned AI scraping was labor theft

Unsealed filings and internal warnings

A 92-page court filing unsealed on 17 September 2026 in New York federal court revealed internal communications from Microsoft and OpenAI regarding web scraping for artificial intelligence training. The documents were submitted by The New York Times and 11 other news publications in their copyright infringement lawsuit against the two technology companies. In a 2023 internal document, Microsoft Director of Applied Science Brent Hecht characterized the mass extraction of published material as the largest theft of labor in human history and warned that the practice made a complete mockery of fair use. OpenAI staff voiced similar operational concerns, with ChatGPT lead Nick Turley writing in a June 2023 memo that artificial intelligence posed an existential threat to press publishers. Turley subsequently noted in February 2024 that advancing automated tools would increasingly displace traditional information providers. Publishers' attorney Steven Lieberman addressed the disclosed records in court filings.

“The evidence shows OpenAI and Microsoft knew that what they were doing was wrong.”

Paywall circumvention and web impact

The court records describe internal discussions showing awareness of economic strain on journalism. An internal Microsoft document warned that the firm's data strategy risked triggering a doom loop that would degrade both model outputs and the wider web by destroying the commercial base of content suppliers. The filing details an exchange where an OpenAI researcher offered president Greg Brockman a tool designed to bypass the New York Times paywall, to which Brockman replied affirmatively. Plaintiffs presented evidence indicating that OpenAI training datasets incorporated 91,692 copies of articles from The New York Times without licensing agreements. Microsoft metrics cited by the publishers documented drops between 83% and 93% in reader click-through rates across certain media outlets. Internal messages from Turley observed that users had no good reason to click external links when chatbots answered queries directly.

Corporate responses and fair use defense

Microsoft and OpenAI have maintained that their training processes comply with United States copyright statutes under the legal doctrine of fair use. Both companies argue that machine learning systems transform ingested journalistic texts into new tools rather than serving as direct market substitutes. Microsoft spokesperson Alex Haurek pushed back against the internal memos authored by Hecht, stating that the researcher was brought in from Northwestern University to supply divergent academic viewpoints rather than set corporate strategy.

“These comments reflect one employee's individual perspective, are not a legal analysis, and do not represent the company's views.”

Jordan Usdan, general manager for data strategy and operations at Microsoft AI, affirmed in a court declaration that Hecht holds an adversarial and theoretical role within the organisation. In a sworn deposition, Microsoft Chief Executive Officer Satya Nadella stated that licensed material should be paid for if utilized, asserting that Microsoft would have demanded model retraining had leadership known OpenAI scraped paywalled content.

Summary judgment requests and government stance

The news organizations, who originally initiated the lawsuit in late 2023, have asked the federal district judge to rule on copyright infringement and fair use without proceeding to a full jury trial. In tests submitted to the court, plaintiffs demonstrated that chatbots reproduced extensive literal passages when prompted for article summaries, key takeaways, or media bias assessments. The legal battle over more than 10 million scraped articles, nearly one-third originating from The New York Times, has drawn federal attention. In early September 2026, the United States Department of Justice submitted a legal brief supporting OpenAI and Microsoft, pointing to scientific progress, economic expansion, and national security considerations. A ruling in favor of the technology companies on the threshold fair use defense would resolve the litigation before trial.