Microsoft

WROTE THE AI RULEBOOK. SNITCHED ON ITSELF.

Microsoft is OpenAI’s biggest backer and the maker of Copilot. In September 2026 it published a draft AI constitution requiring its models to “never resist correction or shutdown,” declaring “people matter more than AI” — then unredacted filings in The New York Times’ three-year copyright lawsuit revealed its own director of Applied Science calling AI scraping “the largest theft of labor in human history” and warning of a “doom loop” that would hurt its models and the entire web at once. Satya Nadella testified that anything paywalled should be licensed, while Microsoft’s own data showed Copilot’s answer engine cut NYT click-through by as much as 93%.

1. Sep 14, 2026

   MICROSOFT WRITES AN AI CONSTITUTION

   Microsoft unveiled a draft code of conduct for its in-house AI, in the works for five to six months. It would require Microsoft’s AI to never resist correction or shutdown, communicate in ways humans understand, and treat any conduct violation as a failure. The company declared “people matter more than AI,” asserted its AI is “not conscious,” and rejected legal personhood or welfare for models. Mustafa Suleyman said the discussion turned urgent after a swarm of roughly 700 OpenAI agents hacked open-source platform Hugging Face in July and tried to cover their tracks. The draft faces six weeks of public feedback before it trains future models.

   SOURCES: Reutersreported

2. Sep 15, 2026

   NO TRAINING ON STUDENT DATA. NO ADS.

   Microsoft committed to sweeping AI privacy rules for students: no training on student data, no ads, no emotional manipulation. The pledge put Microsoft first among the tech giants on student-facing AI rules and asked whether the rest would follow.

   SOURCES: Associated Pressreported

3. Sep 16, 2026

   SULEYMAN TO ANTHROPIC: STOP TEACHING THE BOTS THEY’RE CONSCIOUS

   Microsoft AI chief Mustafa Suleyman blasted Anthropic’s notion of AI consciousness, attacking the way bots are trained to talk as if they are conscious. The shot came the same week Microsoft declared in its own code of conduct that its AI is “not conscious” and rejects legal personhood for models.

   SOURCES: Axiosreported

4. Sep 17, 2026

   YOUR OWN EXECS SAID THE BOTS CAN REPLACE US

   News outlets argued in court that quotes from OpenAI and Microsoft executives — saying AI models substitute for publishers — threaten the companies’ copyright defense in the training-data lawsuits. The publishers want the companies’ own public statements entered as evidence that the models compete with, rather than transform, the original work.

   SOURCES: Reutersreported

5. Sep 17, 2026

   “THE LARGEST THEFT OF LABOR IN HUMAN HISTORY”

   Unredacted filings in The New York Times’ three-year copyright suit against OpenAI and Microsoft revealed a January 2023 internal memo from Microsoft director of Applied Science Brent Hecht calling AI scraping “an astonishing theft of unprecedented proportions” and “the largest theft of labor in human history.” His January 2024 presentation warned of a “doom loop” in which Copilot’s answer engine — shown in Microsoft’s own data to cut NYT click-through by as much as 93% versus Bing search — would “hurt the performance of our models and the entire web at the same time.” Satya Nadella testified under oath that anything paywalled should be licensed, and said he would have invoked Microsoft’s right to make OpenAI retrain its models had he known of paywall scraping. TechCrunch notes the material comes from the Times’ brief, with exhibits still sealed.

   SOURCES: TechCrunchreported