Microsoft Calls AI Scraping ‘Theft’ in Private Emails

The update

Unsealed court documents from a lawsuit involving The New York Times show Microsoft executives privately describing AI data scraping as “the largest theft of labor in human history.” Internal emails from Microsoft’s Director of Applied Science, Brent Hecht, and OpenAI leadership suggest they viewed the practice as a threat to news organizations rather than a legal defense. The documents reveal Microsoft’s Copilot “answer engine” caused click-through rates for news sites to drop as much as 93% compared to traditional search.

Why it matters

The documents contradict the companies’ public arguments that training AI on news content is fair use. Microsoft data cited in the filings shows a “doom loop” where AI products reduce traffic to news sites, potentially damaging the very content supply chains that power these models. OpenAI’s ChatGPT head Nick Turley warned internally that publishers would face an “existential threat” from commercial products trained on news content.

What to watch

Legal experts are watching whether these internal admissions will impact the ongoing copyright cases. The revelation also raises questions about how AI companies will address the economic impact on publishers as they scale their products, particularly as courts have been largely favorable to AI companies’ “fair use” arguments.

Sources

• arstechnica.com — Core reporting on internal Microsoft and OpenAI emails and the 'doom loop' concept.
• techcrunch.com — Context on the lawsuit, click-through rate drops, and the 'existential threat' quote.