In This Article
Unsealed filings from The New York Times’ copyright case against OpenAI and Microsoft reveal that senior executives privately characterized AI scraping as “the largest theft of labor in human history.” The documents also describe paywall circumvention, extensive copying, and concerns that generative AI could undermine the publishers whose work supplied its training data.
Key takeaways
- Microsoft’s Brent Hecht reportedly described AI training as an “astonishing theft” in a 2023 memo.
- OpenAI leaders acknowledged that chatbots could pose an “existential threat” to publishers.
- Internal data showed Microsoft Copilot reduced click-throughs to The New York Times by as much as 93% versus traditional Bing search.
- The filings allege paywall bypassing, mass scraping, and removal of copyright notices.
The disclosures add pressure to a case that could help define how copyright law applies to training large language models. However, much of the newly public information comes from The Times’ legal brief rather than complete underlying exhibits, meaning several statements lack their original context.
What the filings allege
According to the filing, OpenAI and Microsoft assembled large training datasets containing copies of news organizations’ work. OpenAI’s mid-training data allegedly included more than 91,692 works from The Times, Daily News, and the Center for Investigative Reporting. A Common Crawl-derived dataset reportedly contained more than two million documents from nytimes.com.
The filing also describes collaboration between the companies. OpenAI allegedly provided Microsoft with the complete GPT-3 training dataset, while Microsoft supplied data through initiatives known as Project Taxi and Project Mango. The latter reportedly contained at least 160,903 distinct works from news publishers.
Executive comments challenge the fair-use defense
The statements could undermine arguments that AI training is transformative and does not harm the market for original works. Microsoft’s Hecht reportedly warned that generative AI could disrupt the employment of the people who created the training data. OpenAI executive Nick Turley allegedly wrote that publishers faced an existential threat because chatbots were increasingly substitutive.
Microsoft CEO Satya Nadella said in a deposition that paywalled content should be licensed for training or grounding. He also testified that Microsoft could have required OpenAI to retrain models if it had known that paywalled material was scraped.
The commercial impact on publishers
Internal Microsoft analysis reportedly found that Copilot’s answer-engine experience reduced visits to The Times’ website by up to 93% compared with traditional Bing. Hecht described a feedback loop in which fewer publisher visits could weaken the content ecosystem that AI systems depend on.
For enterprise technology leaders, the dispute highlights a broader governance issue: data used to power analytics and AI must be traceable, licensed where required, and managed with clear accountability. Organizations building Microsoft Fabric and Power BI environments should apply the same discipline to data lineage, access controls, retention, and source validation. Spargent’s focus on governed data platforms and measurable reporting outcomes reflects why these controls matter: trusted inputs reduce legal exposure, rework, and costly reporting bottlenecks.
What happens next
The court has not resolved whether copyrighted material may lawfully be used to train AI models, and the disclosed claims are contested litigation arguments rather than final findings. OpenAI and Microsoft had not publicly responded to requests for comment in the reporting reviewed.
The case is expected to continue toward a potential trial decision in 2027. Its outcome could affect licensing models, publisher economics, AI product design, and enterprise procurement standards. For businesses adopting AI, the practical lesson is immediate: document where data comes from, confirm permitted uses, and measure whether automation creates value without undermining the information supply chain it relies on.
- Microsoft exec called AI scraping ‘the largest theft of labor in human history,’ new unredacted filings reveal, TechCrunch.
- Microsoft Director Privately Admitted AI Was the "Largest Theft of Labor in Human History," Unsealed Court Documents Show, Futurism.
- A Microsoft executive described AI-powered data scraping as ‘the greatest labor theft in human history.’, GIGAZINE.
