← Front pageReports · Watch · Archive
DRASKER

DRASKER REPORT

published · Updated November 13, 2026

Microsoft and OpenAI Staff Doubts on AI and Publishing Surface in Unsealed Files

The Times' account arrives alongside a run of amicus briefs from publishers, talent agencies and lawmakers pressing the court to reject fair use, a Justice Department statement of interest arguing the opposite, and a voluntary industry opt-out token that would not reach material already collected.3,7,8

KEY POINTS
  • The Times reported that unsealed filings in its suit against OpenAI and Microsoft show internal concern at both companies about training on news articles, including a 2023 Microsoft document and memos by Nick Turley, who led the ChatGPT team at OpenAI.10
  • Turley's June 2023 memo described AI as an 'existential threat' to publishers; by February 2024 he wrote that AI products 'will get more and more substitutive as they get better.'10
  • A separate account by TechRadar, carried by Yahoo, attributes a near-identical 'largest theft of labor in human history' line to a named Microsoft director, Brent Hecht, while the Times describes unnamed employees debating it.2,10
  • Rightsholders, talent agencies and three Democratic lawmakers urged the court to reject fair use, while the Justice Department filed a statement of interest on Sept. 1, 2026 arguing that training on copyrighted texts qualifies as fair use.1,3,4
  • Google, OpenAI, Microsoft and Anthropic pledged to honor a single robots.txt token as an AI-training opt-out, but reporting describes it as prospective and voluntary, without enforcement, and silent on content already collected.2,7
WHAT WE KNOW

The New York Times reported on Sept. 17, 2026 that documents unsealed in its copyright lawsuit against OpenAI and Microsoft describe internal concern at both companies about the use of millions of news articles to train AI systems.10

According to the Times' account of those filings, a 2023 internal Microsoft document said millions of people would soon consider large models 'hoovering up' their work 'an astonishing theft of unprecedented proportions.'10

The Times reported that the documents surfaced at the summary-judgment stage of the case before U.S. District Judge Sidney H. Stein in the Southern District of New York.10

What the unsealed filings are reported to contain

The Times' Sept. 17 account is the origin of the internal-document material now circulating. The newspaper reported that the unsealed records, filed in the Southern District of New York before Judge Sidney H. Stein, show concern inside both Microsoft and OpenAI over the use of millions of news articles to train AI systems.10

Per that account, a 2023 Microsoft document said millions of people would soon regard large models 'hoovering up' their work as 'an astonishing theft of unprecedented proportions,' and Microsoft employees debated whether OpenAI's practice amounted to the 'largest theft of labor in human history' and could create a 'doom loop' threatening model quality.10

The Times also reported a June 2023 memo from Nick Turley, who led the ChatGPT team at OpenAI, describing AI as an 'existential threat' to publishers, and a February 2024 memo in which he wrote that AI products 'will get more and more substitutive as they get better.'10

Context that bears on how far these documents go: the Times itself alleges that OpenAI and Microsoft used its journalism without permission to develop competing products, and those allegations still require legal resolution.3

One quote, two attributions

The most-quoted line from this material has two versions. The Times describes Microsoft employees debating whether OpenAI's practice was the 'largest theft of labor in human history.'10

TechRadar, in reporting carried by Yahoo, states that the unsealed filings show Microsoft director Brent Hecht describing AI scraping in a 2023 internal memo as possibly 'the largest theft of labor in human history.'2

Both accounts trace to the same court record rather than to two independent inspections of the documents, and TechRadar names a specific person where the Times refers to unnamed employees. The quote is therefore reported, not established as Hecht's, pending resolution of whether the two descriptions refer to the same memo.2,10

How the case reached this point

The Times sued Microsoft and OpenAI in December 2023, with both companies invoking fair use, according to Yahoo's account.2

Danielle Coffey, president and CEO of the News/Media Alliance, wrote in a Gazette opinion column that the Times alleged OpenAI trained its models on the newspaper's content without permission and that both companies copied real-time content for use in outputs, and that other publishers filed similar suits.6

Coffey wrote that more than a dozen of the earliest suits, including The New York Times v. Microsoft, are now at the summary-judgment stage.6

IPWatchdog reported that the consolidated MDL combines suits against OpenAI and Microsoft brought by the Times and other news publishers, by Ziff Davis, and by a group of authors led by the Authors Guild seeking class status, with cross-motions for summary judgment on fair use pending before Judge Stein.8

The Times reported that eleven other publishers have joined its lawsuit. Microsoft and OpenAI publicly contend their use of copyrighted material was covered under 'fair use' legal protections.10

Publishers, agencies and lawmakers press the court

Publishers Weekly's Jim Milliot reported that the Association of American Publishers, the Independent Book Publishers Association, the International Association of Scientific, Technical & Medical Publishers and the Copyright Clearance Center jointly filed an amicus brief in In Re: OpenAI Inc. Copyright Infringement Litigation, supporting plaintiffs seeking summary judgment against the fair-use defense.1

That brief, as Publishers Weekly described it, cites an unnamed Microsoft employee who concluded authors were among the economic victims of large language models, 'no longer able to sell their books because there are so many substitutive AI-generated books out there.'1

Publishers Weekly also reported that the brief argues ChatGPT has already generated sequels substituting for George R.R. Martin's forthcoming The Winds of Winter, and described OpenAI's training on A Song of Ice and Fire as a 'research mission' to write the series' final two books.1

AAP president and CEO Maria A. Pallante said in a prepared statement that authors and publishers are 'integral to producing OpenAI's large language models,' and that without guardrails and financial participation the models could diminish the value of copied works and the long-term viability of human authorship, Publishers Weekly reported.1

Separately, IPWatchdog reported that rightsholder briefs included one from the AAP arguing 'Fair use is not a roving license for courts to revise the framework Congress implemented,' alongside filings from William Morris Endeavor Entertainment and from Representatives Jamie Raskin, Jerrold Nadler and Hank Johnson.8

MediaPost's Wendy Davis reported that William Morris Endeavor contended in its amicus brief that its clients are losing paid work and licensing fees to AI output built on their copyrighted works, and that a licensing market for AI use already exists.4

The counter-filings: fair use, licensing and a warning about a 'de facto licensing regime'

The Justice Department filed a statement of interest on Sept. 1, 2026 arguing that using copyrighted texts to train AI models qualifies as fair use, citing innovation, competition and national security considerations, according to remio.ai.3

IPWatchdog reported the same filing date and described the department as urging the court to hold that training LLMs on copyrighted works is fair use, warning that mandatory licensing would favor large technology firms and harm U.S. competitiveness and national security.8

MediaPost reported that former Senator Rick Santorum and former Representative Rich Boucher filed a friend-of-the-court brief in New York Times Co. v. OpenAI arguing that training AI on copyrighted news articles is fair use, and warning that a contrary ruling would impose a de facto licensing regime.4

The publishers answered that OpenAI's position would let defendants freely copy entire news articles to keep users on AI platforms at the expense of publishers, MediaPost reported.4

IPWatchdog reported that OpenAI published a post titled 'Fair use benefits everyone' featuring excerpts from 20 filings, nearly all amicus briefs in the In re: OpenAI, Inc. Copyright Infringement Litigation MDL, urging the Southern District to find that training LLMs on copyrighted works is fair use.8

OpenAI has publicly defended applying fair use to AI training and highlighted briefs supporting its position, Enstarz reported.5

Two accounts of the CAA and Blanchett filings, kept apart

Two published accounts describe creative-industry filings in the same consolidated litigation and do not line up on date, procedure or signatories.5,9

Mezha, citing Deadline, reported that CAA and other signatories filed an amicus brief on October 20 in the consolidated copyright infringement case involving OpenAI, The New York Times, The Intercept and other outlets, arguing that AI licensing at scale is feasible and that AI's future should be built with artists on consent, control and compensation.9

As Mezha described it, the filing states that 'technical silence is not legal consent' and concludes that 'Licensing at scale is not inherently impossible,' disputing claims that licensing deals are prohibitively costly. Mezha reported it was signed by Robert De Niro, Helen Mirren, Julia Roberts, Scarlett Johansson and CAA chief Bryan Lourd, and filed by Bass, Berry & Sims PLC for CAA and other creative-industry organizations.9

Enstarz, on MSN, reported instead that Cate Blanchett and a coalition including the Human Consent Foundation, Creative Artists Agency, the Music Artists Coalition and the National Association of Voice Actors asked a federal court on Oct. 30 for leave to file a brief on AI licensing in the same consolidated litigation.5

The proposed brief, as Enstarz described it, argues that existing rights-clearance and collective licensing systems can serve as a starting point for negotiating AI uses, and states that the coalition takes no position on the ultimate disposition of the summary-judgment motions.5

Enstarz, quoting a statement it attributes to Blanchett via TheWrap, reported her saying: 'If someone wants to use my face, my voice or my image, they ask me first.'5

TIMELINE
  1. The New York Times sues Microsoft and OpenAI; both invoke fair use, Yahoo reported.2

  2. Nick Turley, who led the ChatGPT team at OpenAI, writes in a memo that AI poses an 'existential threat' to publishers, per the Times' account of unsealed filings.10

  3. An internal Microsoft document says the public will soon regard models 'hoovering up' their work as 'an astonishing theft of unprecedented proportions.'10

  4. Turley writes that AI products 'will get more and more substitutive as they get better,' according to the Times.10

  5. The Justice Department files a statement of interest arguing that training AI on copyrighted texts is fair use.3,8

  6. The New York Times reports on unsealed filings in its suit describing internal Microsoft and OpenAI concern over training on news articles.10

  7. Mezha, citing Deadline, reports that CAA and other signatories filed an amicus brief arguing AI licensing at scale is feasible.9

  8. A submission from publishing organizations including the AAP, book and scientific publishers, independent publishers and the Copyright Clearance Center argues that licensing should be central to the copyright analysis, remio.ai reported.3

  9. Representatives Jerrold Nadler, Jamie Raskin and Hank Johnson announce a court submission opposing the Justice Department's fair-use position, remio.ai reported.3

  10. Enstarz reports that Cate Blanchett and a creative-industry coalition including CAA asked a federal court for leave to file an AI licensing brief.5

WHAT WE STILL DON'T KNOW
  • Whether the unsealed filings themselves have been obtained or reproduced; the bundle contains only news reports about them, so the internal documents cannot be quoted as inspected primary sources.
  • Whether the NYT 'employees debated' language and the Yahoo/TechRadar attribution to named Microsoft director Brent Hecht refer to the same internal memo, and whether Hecht can be named on the record.
  • Whether the CAA amicus brief described by Mezha/Deadline (filed Oct 20) and the Blanchett-led coalition leave request described by Enstarz/MSN (Oct 30) are one filing or two, and which date and signatory list is correct.
  • Whether any whistleblower account exists; none appears in the reporting Drasker has located, and the assignment specifically asks about this.
  • Whether Microsoft or OpenAI have issued any on-record response to the NYT reporting on their internal documents.
  • The identity and status of the unnamed Microsoft employee quoted in the AAP amicus brief, and whether that quote reflects an internal analysis or an informal remark.
  • Whether the unsealed filings themselves have been obtained or reproduced; the material gathered contains only news reports about them, so the internal documents cannot be quoted as inspected primary sources.
  • Whether the Times' 'employees debated' language and the TechRadar attribution to named Microsoft director Brent Hecht refer to the same internal memo, and whether Hecht can be named on the record.
  • Whether the CAA amicus brief described by Mezha/Deadline (filed Oct. 20) and the Blanchett-led coalition leave request described by Enstarz/MSN (Oct. 30) are one filing or two, and which date and signatory list is correct.
  • Whether any whistleblower account exists; none appears in the material gathered, and the assignment specifically asks about this.
  • Whether Microsoft or OpenAI have issued any on-record response to the Times reporting on their internal documents.
SOURCES
  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.