AnswersPublished18 sources
Unsealed NYT v. OpenAI Filings Put AI’s News-Copyright Defense Under Pressure
Newly unsealed filings bolster publishers’ claim that OpenAI and Microsoft used news at scale while recognizing that AI answers could substitute for publisher traffic; Microsoft testing cited by the publishers found C... The documents matter most because they could bear on fair use’s purpose and character and market...
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Newly unsealed filings bolster publishers’ claim that OpenAI and Microsoft used news at scale while recognizing that AI answers could substitute for publisher traffic; Microsoft testing cited by the publishers found C...
- The documents matter most because they could bear on fair use’s purpose and character and market effect questions, while the Justice Department has separately argued that AI training generally can qualify as fair use.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What do the newly unredacted filings in The New York Times’ copyright lawsuit against OpenAI and Microsoft reveal about the companies’ alleg. Article summary: The unredacted materials strengthen the publishers’ factual narrative: they portray OpenAI and Microsoft as copying news at industrial scale, allegedly circumventing access controls, and privately recognizing that chatbo. Topic tags: general, general web, user generated, news. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
The newly unsealed materials do not decide The New York Times’ copyright case against OpenAI and Microsoft. But they give the publishers a more detailed evidentiary narrative: news content was allegedly gathered at enormous scale, including from access-controlled sources, and company personnel understood that AI answer products could weaken the publishers’ audience and revenue base. 5
That combination is consequential because the dispute is no longer only about whether models trained on news. It is also about how the material was obtained, what the companies knew about the commercial consequences, and whether those facts affect a fair-use defense.
What the publishers say the records show
In their unsealed filing, the publisher plaintiffs contend that OpenAI and Microsoft copied and used news works at industrial scale. The Times’ original complaint alleged that millions of Times works were copied and ingested, in some instances repeatedly, to train GPT models. 29
The later filing, as reported from the unsealed materials, adds allegations involving large datasets of publisher documents and substantial numbers of distinct copyrighted works. The publishers also allege that the companies sought access to paywalled reporting and removed copyright-management information such as ownership notices. 2
5
Those are allegations, not established facts. Still, they matter to the plaintiffs’ theory because the means of acquisition and treatment of rights information could shape how a court evaluates the commercial character and fairness of the use.
The internal warnings are powerful evidence—but not legal admissions
A central disclosure is attributed to Microsoft Director of Applied Science Brent Hecht, who reportedly described AI scraping as “an astonishing theft of unprecedented proportions” and potentially “the largest theft of labor in human history.” 5
13
The unsealed materials also reportedly include concern that generative-AI products could create a harmful feedback loop for publishers: systems trained on journalism could answer users directly, reducing visits to the outlets that fund reporting. 10
11
For the publishers, the value of these statements is not that an executive used the word “theft.” An employee’s language does not determine copyright liability. Rather, the publishers can use the records to argue that decision-makers and researchers foresaw a specific market-substitution problem.
Microsoft has said that the internal materials reflect individual views rather than legal conclusions and has defended Copilot’s uses as transformative. 11
Why the reported 93% click-through figure stands out
The filing reportedly cites Microsoft testing showing that Copilot could reduce click-throughs to nytimes.com by as much as 93% compared with ordinary Bing search. 4
That figure is particularly relevant to the fourth fair-use factor: the effect of the use on the potential market for or value of the copyrighted work. The publishers’ argument is straightforward: if an answer engine satisfies the reader’s need without sending them to the original article, it may function as a substitute rather than as a referral channel.
A single internal test would not, by itself, prove market harm across all queries, products or publishers. But it gives the plaintiffs a concrete basis for arguing that the alleged harm is measurable and foreseeable, rather than speculative.
Paywalls and copyright notices could complicate the fair-use fight
OpenAI and Microsoft’s broad position is that model training can be transformative fair use. The U.S. Justice Department has supported that general proposition in a statement of interest, urging the Manhattan federal court to reject a rule that would broadly treat training on copyrighted text as infringing. 33
The government’s intervention is meaningful, but it is not a judicial ruling and does not resolve every allegation in this litigation. The court could distinguish between the general act of training a model and evidence, if proven, of bypassing paywalls, stripping rights information, producing infringing outputs, or deploying a product that displaces a licensing or readership market.
In other words, the case may turn less on an abstract question—whether AI training can ever be fair use—than on the particulars of the alleged conduct and commercial effects.
What the filings change
The disclosures strengthen the publishers’ litigation story in three ways:
- They add alleged evidence of scale. The plaintiffs can frame the copying as systematic rather than incidental.
29
- They add alleged evidence of intent and knowledge. Internal concerns can support the argument that the companies understood the risks to journalism and publisher economics.
5
11
- They add an alleged measure of substitution. The reported Copilot click-through analysis gives the market-harm argument a specific data point.
4
None of that automatically defeats fair use. Copyright analysis remains fact-specific, and the court has not ruled that the alleged training, scraping or product behavior was unlawful. The Justice Department’s position also ensures that OpenAI’s broader fair-use argument has influential support. 33
The central question now is sharper: if the plaintiffs can prove that protected, paywalled journalism was acquired and used in ways that replaced visits and licensing opportunities, will a court view that as transformative innovation—or as a commercially harmful use that fair use does not protect?
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Unsealed NYT v. OpenAI Filings Put AI’s News-Copyright Defense Under Pressure"?
Newly unsealed filings bolster publishers’ claim that OpenAI and Microsoft used news at scale while recognizing that AI answers could substitute for publisher traffic; Microsoft testing cited by the publishers found C...
What are the key points to validate first?
Newly unsealed filings bolster publishers’ claim that OpenAI and Microsoft used news at scale while recognizing that AI answers could substitute for publisher traffic; Microsoft testing cited by the publishers found C... The documents matter most because they could bear on fair use’s purpose and character and market effect questions, while the Justice Department has separately argued that AI training generally can qualify as fair use.
Sources
gigazine.netA Microsoft executive described AI-powered data scraping as 'the greatest labor theft in human history.'
washingtonpost.comMicrosoft exec called AI the ‘largest theft of labor’ in history, court records show