Newly unsealed court documents show a Microsoft executive calling his own company's AI strategy "the largest theft of labor in human history."
The filings, unsealed this week in The New York Times' copyright suit against OpenAI and Microsoft, quote Microsoft's director of applied science, Brent Hecht, warning internally that Copilot's answer engine could trigger a "doom loop" for the web. Hecht described AI systems scraping journalism for its value, then repackaging it into instant answers that undercut the sites they took it from. Microsoft's own data showed click-through rates to Times search results falling 87 to 93 percent versus standard Bing searches, with rates for Ziff Davis sites like IGN and Eurogamer dropping 51 to 94 percent. The filings also show OpenAI president Greg Brockman replying "ah nice" when a researcher described a way around the Times' paywall for training data.
This isn't a hypothetical harm anymore - it's Microsoft's own applied-science team documenting, in writing, that the product guts the publishers supplying it. Fair-use arguments tend to lean on AI answers being transformative; internal click-through numbers this stark make that a harder sell in court. Satya Nadella testified he would have made OpenAI retrain its models had he known paywalled content was involved, which reads as either a real limit or a convenient one to state after the fact.
Executives rarely put "doom loop" and "theft of labor" in writing about their own products - that's usually left to critics outside the building.