OMITTED

What the news leaves out.

← Omitted front page

Hackers Breached OpenAI Adding to Safety Concerns

6 sources · updated 2026-09-19
Left 100% Center 0% Right 0%
2 left · 0 center · 0 right

What happened

In 2023, The New York Times sued OpenAI and Microsoft in the U.S. District Court for the Southern District of New York, alleging the companies trained generative-AI models on its journalism without permission or payment; other publishers later joined, and Judge Sidney H. Stein is considering summary-judgment motions. Newly public litigation materials — much of the language appears in The Times’s brief, with many underlying exhibits still sealed — quote Microsoft director of applied science Brent Hecht calling mass scraping of news “the largest theft of labor in human history” and warning that the firms’ AI content strategy had started a “doom loop” that would “hurt the performance of our models and the entire web at the same time.” OpenAI ChatGPT lead Nick Turley wrote that AI posed an “existential threat” to publishers and that products “will get more and more substitutive as they get better,” and the filings cite Microsoft data showing Copilot cut click-through rates to The Times’s domain by as much as 93 percent compared with traditional Bing search. The materials also describe training data containing tens of thousands of copies of plaintiff works, a researcher telling OpenAI president Greg Brockman of a “hack to get around nytimes paywall” (Brockman replied “ah nice”), and Microsoft CEO Satya Nadella’s testimony that “anything that is paywalled should be licensed” and that he would have required OpenAI to retrain models had he known they used paywalled content. Microsoft said Hecht’s comments reflected one employee’s individual perspective, were not a legal analysis, and did not represent the company’s views, which remain that the uses are transformative and consistent with copyright law.
Sourcing check · Mostly on the record · 4 articles

3 of 4 articles cite a named source for the story. It was on the record from the first report. Microsoft spokesperson denied part of it on the record.

What these terms mean and how they're counted
On the record · names a source 3 of 4
On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Brent Hecht, Satya Nadella and Sam Altman
Primary evidence · cites a document or data 3 of 4
Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: court filing
Credits another outlet 0 of 4
The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
Unnamed sources 0 of 4
The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
On-record denial 1
A named person or institution says the claim, or part of it, is not true. It is on the record — someone signed their name to it — but it is not confirmation, so it never counts toward On the record. Criticizing or dissenting from a decision is not a denial. The count is how many people or institutions deny it; each is listed with the first article that carried the denial. Here: Microsoft spokesperson (New York Post, Sep 18, 14:20 UTC)
How it developed
When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Sep 18, 01:12 UTC by New York Times, naming its source. First document in the first report, from New York Times (court filing).
Not counted 0
Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out

Unpacked

Right-leaning coverage reports Microsoft's own measure of how sharply its AI answers cut newspaper clicks; left-leaning coverage never gives that figure. The New York Post says click-through rates for the Times fell as much as 93 percent when Bing's AI version was compared with ordinary search. Left-leaning accounts stay with staff warnings of a doom loop and an existential threat. Those memos then read as unease. The missing number is the companies' evidence that the product was already taking the click. The Verge says the companies "did it anyway." The New York Post says workers "privately shared concerns," then records the firms' public claim that the models turned the articles into new content. Did anyone at Microsoft require a retrain after its CEO said paywalled work should be licensed?
Bottom line

The Times' own write-up of the unsealed filings left out Microsoft's finding that click-through rates fell as much as 93 percent with AI search, a figure the New York Post put in print.

The receipts — 4 facts one side's coverage lacks
  • Microsoft measured an 83% to 93% drop in click-through rates for the New York Times and Daily News when comparing Bing's AI version with traditional search numberin New York Post · absent from left-leaning coverage
  • OpenAI employees said GPT-4 memorized a ton of data and would be insanely good at regurgitation, and the filing cites verbatim article outputs contextin The Verge · absent from right-leaning coverage
  • Brent Hecht wrote that Microsoft's defense makes a complete mockery of the idea of fair use contextin The Verge · absent from right-leaning coverage
  • An OpenAI representative said he was unaware of any effort to detect or remove paywalled content from training data official statementin The Verge · absent from right-leaning coverage
The Left View
Left-leaning outlets treated the filings as evidence that OpenAI and Microsoft understood the harm and proceeded anyway. The Verge called the documents “pretty damning,” said the companies “knew they were driving us toward Google Zero, and did it anyway,” and highlighted Hecht’s claim that the defense made a “complete mockery of the idea of ‘fair use,’” internal language that LLMs are “a product that destroys its own supply chain,” and Greg Brockman’s reported interest in “gazillions” of dollars from commercial AI. The New York Times and The Verge stressed alleged paywall workarounds, employees’ admission that GPT-4 would be “insanely good at regurgitation” of copyrighted text, and OpenAI experts linking AI summaries to collapsing referral traffic — framing the episode as firms knowing they would damage “millions of people” in publishing, and even their own future data supply, yet continuing “doom loop be damned.”
The Right View
Right-leaning coverage reported the same internal warnings but gave more weight to the companies’ public legal position and Microsoft’s effort to cabin Hecht. The New York Post noted that Microsoft and OpenAI have defended scraping as lawful because articles were “turned into completely new content,” and quoted Microsoft saying Hecht’s remarks “are not a legal analysis, and do not represent the company’s views,” that “transformative uses are consistent with copyright laws,” and that Copilot “is not a substitute for publishers’ journalism.” The Post and Washington Examiner also relayed Nadella’s licensing line on paywalled material and Microsoft’s claim that his testimony is “perfectly consistent” with its court filings, while still headlining the “largest theft of labor” and “doom loop” language and placing the fight in the context of publishers seeking damages and rushing into licensing deals as traffic falls.
Our Take (balanced)
The left’s strongest argument is the paper trail of substitution and harm: Hecht’s “theft” and “doom loop” language, Turley’s “existential threat” and “substitutive” warnings, Microsoft’s own click-through collapse, documented regurgitation, and the paywall-hack exchange all go to whether chatbot answers replace the market for the original journalism. The right-leaning coverage’s strongest counter is institutional rather than factual denial: Microsoft has disowned Hecht as an academic outlier, the companies still plead transformative fair use, and Nadella testified he would have forced a retrain over paywalled scraping. The unresolved tension is whether those internal descriptions of market destruction count as corporate knowledge that defeats a fair-use defense, or as recorded dissent that does not change the legal character of training on copyrighted news.

6 sources

More stories

The week's bottom lines, in your inbox

One email a week: the five stories that mattered and what they actually mean. Free.