Microsoft and OpenAI Files Expose AI News Scraping Warnings in Copyright Fight
Unsealed filings show Microsoft and OpenAI insiders warning that AI news scraping threatens publishers, as plaintiffs seek to defeat fair use legal claims.
Summary
On September 17, 2026, a summary judgment motion unsealed Thursday by plaintiffs led by The New York Times exposed Microsoft and OpenAI warnings predating ChatGPT and Copilot. Microsoft scientist Brent Hecht called mass news scraping potentially history’s largest labor theft and a mockery of fair use, saying almost no creators intended or received payment for it. ChatGPT head Nick Turley predicted an existential threat and called chatbots largely substitutive; a Microsoft document warned of a doom loop damaging publishers, model quality and the web.
Microsoft measured click-through declines of 83 to 93 percent for some plaintiffs and 51 to 94 percent for others. Under oath, CEO Satya Nadella said AI companies should not evade paywalls and acknowledged chatbots keep users from underlying sources. OpenAI staffer Nick Ryder told President Greg Brockman of a hack around The New York Times paywall; Brockman replied, “Ah, nice.” Another engineer and Turley said prominent links would not produce clicks. Tests generated long or verbatim excerpts through summaries, bullet points, bias ratings and homepage requests. Plaintiffs seek judgment only for articles with extensive overlap, reserving other claims for trial.
Plaintiffs also allege Microsoft built a filter Hecht feared could resemble an accidental cover-up, repurposed data bought for Bing as OpenAI training material without publisher consent, and rejected licensing. OpenAI allegedly used a third-party New York Times dataset containing 1.8 million articles despite noncommercial restrictions and employee concerns. They say ChatGPT encouraged Google AI Overviews and wider free-riding; Brockman cited the “gazillions” available from commercialization. Microsoft says its products are transformative fair use and Hecht expressed a personal view. Steven Lieberman, counsel for the New York Daily News and seven sister papers, alleges knowing wrongdoing. Publishers are preparing for trial and seek licensing rules.
Positives
- September 17, 2026 filings brought previously confidential Microsoft and OpenAI discussions into public court proceedings.
- Satya Nadella testified that AI companies should not bypass news paywalls or violate publishers’ terms of use.
- Plaintiffs limited their summary judgment request to articles showing extensive verbatim overlap, leaving broader claims for trial.
- A court requirement to license news could place Microsoft, OpenAI and rival AI companies under the same compensation rules.
Risks & concerns
- Microsoft measured click-through declines of 83 to 93 percent for some news plaintiffs and 51 to 94 percent for others.
- Brent Hecht warned that uncompensated AI scraping could represent history’s largest theft of labor and undermine fair use.
- Nick Turley described chatbots as largely substitutive and predicted an existential threat to commercial news providers.
- OpenAI allegedly used 1.8 million restricted New York Times articles despite employee concerns about training on the dataset.
- Nick Ryder identified a crawler hack around The New York Times paywall, prompting an approving response from Greg Brockman.
- Microsoft documents warned that declining publisher revenue could trigger a doom loop that weakens journalism, AI models and the web.