Papalocal
Loading…
News
← All news
Technology

Microsoft exec privately called AI news scraping 'largest theft of labor in human history'

Unsealed court filings in the NYT's copyright lawsuit reveal that Microsoft and OpenAI executives warned internally that their own AI products posed an 'existential threat' to the news industry — even as they built them.

How does this story make you feel?

In January 2023, as ChatGPT was reshaping the internet, a senior Microsoft scientist sat down and wrote what amounted to an internal indictment of his own company's practices. Brent Hecht, Microsoft's Director of Applied Science, called the mass scraping of news content to train AI models 'an astonishing theft of unprecedented proportions' — and, in the same memo, 'the largest theft of labor in human history.' That document, long shielded from public view, was unsealed Thursday as part of a motion for summary judgment filed by news organizations led by The New York Times in their copyright lawsuit against Microsoft and OpenAI.

The unsealed filings, reported simultaneously by Ars Technica, TechCrunch, The New York Times, and The Washington Post, pull back the curtain on years of internal deliberations at both companies. Taken together, the documents show executives and engineers who understood — and in some cases explicitly stated — that their AI products would economically devastate the news publishers whose work made those products possible.

Hecht's memo went further than the theft characterization. In a separate document, he wrote that the plan to widely scrape news content made 'a complete mockery of the idea of fair use' — directly contradicting the legal defense that Microsoft and OpenAI have publicly maintained throughout the lawsuit, according to Ars Technica.

It is highly unusual that an end-product threatens the economic foundations of its essential suppliers, but that is the situation we have created for our LLM business with respect to its 'content supply chain.'— Internal Microsoft document, as quoted in the court filing

A Microsoft document outlined what it called a 'doom loop': as AI products siphon traffic away from news sites, publishers lose revenue, produce less journalism, and ultimately leave AI models with less reliable content to draw from — hurting 'the performance of our models and the entire web at the same time,' according to Ars Technica. Microsoft's own data, cited in the filing, showed its Copilot 'answer engine' caused click-through rates for some news plaintiffs to drop between 83 and 93 percent compared to traditional Bing search results, according to TechCrunch.

At OpenAI, the internal picture was similarly stark. In internal communications, ChatGPT head Nick Turley described publishers as facing an 'existential threat' from products like the chatbot. He described the technology as 'largely substitutive, period,' and predicted it would 'get more and more substitutive as they get better,' according to Ars Technica. One of the company's software engineers wrote in an internal message that 'no matter how prominently we show the links, users won't click,' a view Turley echoed, saying there was 'no good reason to click' when the chatbot provides the information directly.

The filings also detail how the companies allegedly obtained the content in the first place. According to TechCrunch, OpenAI's mid-training datasets are said to hold more than 91,692 copies of works from the NYT, Daily News, and Center for Investigative Reporting. More than 2 million documents drawn from nytimes.com alone were found in a Common Crawl-derived dataset. The companies also allegedly ran data-sharing initiatives — internally called Project Taxi and Project Mango — that assembled copies of at least 160,903 unique works from news publishers, per TechCrunch.

The filing also describes alleged efforts to bypass paywalls. When OpenAI researcher Nick Ryder told company President Greg Brockman that a 'hack to get around nytimes paywall' had been found, Brockman replied: 'ah nice,' according to both Ars Technica and TechCrunch. Separately, OpenAI employees allegedly stripped copyright notices from training data before it reached the model, because researchers 'wouldn't want model outputting' those notices to users, according to TechCrunch.

The disclosures put Microsoft CEO Satya Nadella in a complicated position. Under oath, Nadella testified that 'anything that is paywalled should be licensed by anyone who wants to use it for grounding or training,' and said that had he known OpenAI scraped content from behind paywalls, he would have invoked Microsoft's right to require OpenAI to retrain its models, according to TechCrunch. Under oath, he also conceded that chatbots have substituted for news platforms by 'giving you the information right there on the website on the AI platform versus needing to go to the underlying source,' per Ars Technica.

In a Microsoft document, Hecht acknowledged that 'almost no one intended for content they created to be used in this fashion, nor are they compensated for its use,' according to Ars Technica.

The evidence revealed here for the first time shows that OpenAI and Microsoft knew that what they were doing was wrong.— Steven Lieberman, counsel for the New York Daily News

The news plaintiffs are seeking summary judgment — a ruling from the court without a full trial — on articles where AI outputs show 'extensive verbatim overlap' with original reporting. Their motion argues that the companies' own internal statements demolish the fair use defense, particularly the requirement that a use not substitute for or harm the market for the original work. News organizations say they are prepared to go to trial on remaining claims.

The question of whether AI companies can legally train on copyrighted material remains unsettled in U.S. courts, and judges have so far been broadly receptive to fair use arguments from AI firms. Earlier this month, the Trump administration filed a brief in support of OpenAI's position, according to TechCrunch. Microsoft and OpenAI did not respond to requests for comment from TechCrunch. The underlying exhibits in the case remain sealed; the new disclosures come from the news plaintiffs' own brief.

The future not just of journalism but of responsible AI too depends on preserving incentives for humans to produce the creative works on which a healthy society depends.— News plaintiffs, in their court filing

Why it matters — If courts accept that Microsoft and OpenAI's own executives privately acknowledged their AI products constitute theft and will destroy the news industry, it could gut the fair use defense that the entire AI training ecosystem currently rests on.

⚠ Not yet confirmed

  • Much of the new information comes from The Times' own legal brief rather than the underlying exhibits, which remain sealed, meaning the internal quotes appear without their full original context.

Reported by arstechnica.com, techcrunch.com, nytimes.com, washingtonpost.com

People in this story
  • Satya Nadella
    Indian-American business executive and CEO of Microsoft
  • Greg Brockman
    American entrepreneur, investor, and software developer

Community

Be the first to comment.

Share this story
Text Email