Microsoft's own data kills AI's fair use defense — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Microsoft Applied Science director Brent Hecht called AI scraping "the largest theft of labor in human history" in a January 2023 internal memo unsealed this week.link ›
- Microsoft's own data showed Copilot cut click-throughs to nytimes.com by up to 93%, with an internal deck warning of a "doom loop" for both models and "the entire web."link ›
- OpenAI's Nick Turley wrote internally that publishers face an "existential threat" from products "largely substitutive" — language that undercuts the fair use market-harm prong.link ›
- Satya Nadella testified paywalled content "should be licensed by anyone who wants to use it" and said he would have forced OpenAI to retrain had he known it scraped paywalled works.link ›
- OpenAI's mid-training datasets alone contain more than 91,692 copies of works from the NYT, Daily News, and Center for Investigative Reporting.link ›
- A US military strike on a Chinese vessel was aborted this spring after officials discovered the targeting intelligence had been hallucinated by an AI chatbot, first reported by CNN.link ›
- Anthropic plans a November IPO and is preparing a new model to counter GPT-6 Astra, even as Dario Amodei publicly calls for slowing AI development.link ›
The unsealed Microsoft and OpenAI filings in The New York Times's copyright suit just detonated AI's cleanest legal defense. Brent Hecht, Microsoft's director of applied science, called scraping "the largest theft of labor in human history" in a January 2023 memo. Nick Turley, OpenAI's head of ChatGPT, wrote internally that publishers face an "existential threat" from products "largely substitutive" — language that does fatal work on the market-harm prong of fair use. Microsoft's own data showed Copilot slashed click-throughs to nytimes.com by up to 93%, and Satya Nadella testified paywalled content "should be licensed by anyone who wants to use it." Days later, news broke that an AI hallucination nearly triggered a strike on a Chinese vessel this spring. The industry's "trust us" defense is collapsing on two fronts.
The stories behind this week
Microsoft Called AI Scraping 'Largest Theft of Labor'The unsealed filings give The New York Times concrete evidence that the AI companies' own executives viewed scraping as theft and acknowledged products are substitutive — directly undermining the fair use defense that has so far carried the day in court. With Microsoft's own data showing a 93% click-through collapse, publishers gain a quantifiable market-harm figure to anchor damages arguments and potentially licensing demands.
Google opens Google Home MCP to Claude, ChatGPTGoogle is gating this behind its $20/month Premium Advanced tier in the U.S. only — a small audience of paying smart-home enthusiasts gets first crack at wiring frontier AI agents into their homes, while broader availability and device-maker implications remain unaddressed.
AI Hallucination Nearly Triggered US Military StrikeThe near-miss exposes the Pentagon's AI acceleration paradox: the same speed meant to shorten kill chains let hallucinated intelligence reach operational commanders before anyone caught it. Steckler warned there are 'life and death consequences' when AI errors feed targeting decisions — a warning the cargo-manifest hallucination made concrete.
With iOS 27, I’m actually using Siri againApple's decision to rebuild Siri on Google's Gemini models — after a delayed rollout from Apple Intelligence's 2024 tease — marks a tacit acknowledgment that its in-house Siri could not keep pace, giving Google an inroad to potentially hundreds of millions of iPhone users and reshaping the competitive AI assistant landscape.
Anthropic co-founder backs mandatory AI kill switchesAnthropic's own leaders are publicly backing rules that could constrain their models, while Trump calls the underlying fear a 'HOAX' and the UK dismisses a kill switch as pointless. A record Anthropic IPO is reportedly looming while OpenAI delays its $852bn offering over the safety debate — making the policy fight a direct hit on the industry's biggest financial moments.
Valve's Steam Frame Launches at £1,089 Amid VR Cost DebateThe £1,089 price tag lands as RAM prices have more than doubled in a few months, squeezing hardware makers across the industry and forcing Valve to admit its pricing was a compromise rather than a choice. With Meta's Quest subsidies looking unlikely to continue, VR's cheapest path may narrow just as Valve pitches the Steam Frame to enthusiasts willing to pay top tier.
Anthropic Considers New Model Ahead of November IPOAnthropic's November IPO date, set after Amodei's public call for an AI slowdown, exposes the gap between the company's stated safety posture and its aggressive commercialization. Revenue projections of $100B+ this year, up from $65B in July, show investor appetite for that pace is strong.
OpenAI acquires Israeli-founded camera startup Glass Imaging for over $300 million - calcalistech.comA $300M+ acquisition brings OpenAI into direct ownership of smartphone-camera IP at a moment when AI-device makers are racing to own their imaging stacks. The founder's reported Rafael defense background, visible in a cut-off companion headline, suggests the imaging tech pedigree runs deeper than consumer phone cameras — and three major outlets converging on the story within hours signals unusually tight sourcing.
Why it matters: OpenAI and Microsoft's own internal documents now give publishers a paper trail to demand licensing, dismantle the fair use defense, and reset what 'training data' actually costs.
Ask SkimNews
