Dossier mode
“An Astonishing Theft”: OpenAI and Microsoft in Their Own Words
The same investigation, restaged one beat at a time. Drive it with the arrow keys, space, or autoplay. Nothing is cut from the piece — long runs are split across frames. Read the full investigation or open the Who Controls What You Get to Know hub.
“An astonishing theft.”
That's a Microsoft executive's phrase, quoted in the newspapers' case against OpenAI and Microsoft — 'the largest theft of labor in human history.' The publishers' summary-judgment brief is built almost entirely from the defendants' own documents and sworn testimony.
The filing and the admissions it quotes are FACT — on the public record in federal court. Whether the conduct is ultimately ruled infringement or protected fair use is undecided, and graded SOME SMOKE. The words are settled; the verdict isn't.
COI up front: this page was drafted with Claude, made by Anthropic — an OpenAI competitor that settled its own authors' copyright case for a reported $1.5B and destructively scanned books (Project Panama). We grade our maker HARDER, not softer.
Five news organizations are asking a federal court to rule OpenAI and Microsoft infringed at every stage of building their AI.
On Sept 17, 2026, The New York Times, the Daily News papers, Ziff Davis, the Center for Investigative Reporting, and The Intercept filed a combined summary-judgment brief in the consolidated MDL before Judge Sidney Stein (S.D.N.Y., 25-md-3143). It seeks pre-trial rulings that the defendants infringed by acquiring, training on, grounding on, outputting, and 'horse trading' the articles; that fair use fails as a matter of law; that OpenAI intentionally stripped copyright-management information (DMCA); and that damages run per article. Microsoft has put $10B+ into OpenAI for a ~20% revenue share and equity; OpenAI reportedly eyes an IPO near a $1 trillion valuation.
The defendants' own people called it theft and admitted the products substitute for news.
Per the brief, quoting internal docs and depositions: a Microsoft applied-science director called it 'an astonishing theft of unprecedented proportions' and 'the largest theft of labor in human history'; another Microsoft exec said a fair-use win would 'make a complete mockery of the idea of fair use.' OpenAI's head of ChatGPT Nick Turley wrote the products are 'largely substitutive, period' and an 'existential threat'; once a chatbot answers there is 'no good reason to click.' CEO Satya Nadella testified chatbots 'substituted' for visiting publisher sites. A Microsoft memo: the AI strategy started a 'doom loop' threatening 'the economic foundations of its essential suppliers'; internally, large language models are 'a product that destroys its supply chain.' Greg Brockman: the models are 'excellent at news,' and he was 'deeply motivated by the gazillions.' Per the brief, 'Defendants do not dispute that the reason ChatGPT and Copilot are good at news is because they trained on stolen news content' (SF1428).
How the material was taken: paywall workarounds, a non-commercial corpus used anyway, CMI stripping, and 'horse trading.'
Per the filing: OpenAI's scraping 'did not include reviewing websites' Terms of Use or Service' and it had no method to detect paywalled content; told of 'a hack to get around [the] nytimes paywall,' Brockman replied 'ah nice.' OpenAI trained on the New York Times Annotated Corpus despite its non-commercial-only license, after employees said it 'would not be appropriate.' It hosted Custom GPTs named 'Bypass Paywall,' 'Remove Paywall,' and 'NYTimesGPT,' and systematically removed copyright-management information before training (the DMCA claim). Rather than license, the two firms swapped copies in deals they called 'horse trading.' Nadella testified 'anything that is paywalled should be licensed.'
The market-harm numbers — many the defendants' own — show the news business being drained.
Per the brief: Microsoft's own data showed 83–93% drops in click-throughs to Times and Daily News domains (51–94% for Ziff Davis) from its Copilot answer engine vs. Bing. Cloudflare's CEO put OpenAI's pages-scraped-to-visitors-referred ratio at 1,500:1 by mid-2025 (Google 18:1). A study found 87.78% of ChatGPT users visit no external site during search (vs 26.91% for Google). In a survey, 36% of Times subscribers who used ChatGPT for news said it 'means I no longer need news from the New York Times at all.' And the products enable 'pink slime': ~$6,800 in API calls generates a million 500-word news-style articles with no reporter; Prism News ran 200 AI outlets posing as local newsrooms with four employees.
Whether this is unlawful infringement or protected fair use is not yet decided.
This is the plaintiffs' brief — adversarial by design. OpenAI and Microsoft argue training on public material is transformative fair use, that outputs rarely reproduce articles, and that the quotes are selectively framed. The case law is unsettled: Authors Guild v. Google blessed mass copying of books for search, and 2025 rulings in the Anthropic and Meta authors' cases found the act of training can be fair use — while the piracy half of the Anthropic case and Thomson Reuters v. Ross went the other way. A summary-judgment motion is a request, not a ruling; Judge Stein hasn't decided. We grade the admissions FACT and the 'this was theft / fair use fails' conclusion SOME SMOKE.
A machine that eats the thing it needs to work.
The stakes aren't one lawsuit's damages. If AI answer engines divert the readers and revenue that pay for reporting — and the defendants' own 'doom loop' memo says they do — then the machines that summarize the news help kill the newsrooms that make it, and flood what's left with cheap synthetic 'pink slime.' The end state: a public that asks a chatbot 'what's the news?' and gets an answer with no one left to report it. It sits beside Project Panama (the same data appetite led our own maker to destroy books) and The Wrong AI Debate (the industry steering attention to future risks and away from present harms like this one). Control over what you get to know is being transferred from the people who gather facts to the companies that repackage them — litigated in real time.