# Satya Nadella Warns AI Isn't Worth Pursuing Without Human Control > Source: > Published: 2026-09-15 09:25:37+00:00 *Satya Nadella just told his own workforce that Microsoft won't chase superintelligence unless people stay in charge of it.* In an internal memo sent September 14, Microsoft's CEO warned that any AI system not firmly under human control is "not worth pursuing." Nor is one that isn't clearly helping humanity. He called for "deliberate pacing" in frontier AI development. And he backed the idea of embedded evaluators. These are independent monitors stationed inside AI labs, watching how the work actually gets done, not just reviewing it after the fact. Nadella's sharpest line is a business one, not a philosophical one. He told employees that companies neglecting AI safety risk losing their operational license, the practical permission to keep running that regulators and customers grant a business. That's a different kind of warning than "this could go wrong." It's "this could get you shut down." He wants rigorous third-party security testing built into how enterprises deploy AI, not bolted on after a breach. The memo landed the same week Microsoft AI opened public comment on a new Code of Conduct for its in-house MAI models. The 37-page draft, published September 14 and open for review through late October, is built on a blunt premise: people matter more than AI. It bars MAI models from running offensive cyberoperations. It bars them from helping build chemical, biological, radiological or nuclear weapons, and from generating deepfakes. The central rule: the models must never resist correction or being shut down. ## Amodei's warning, and Wall Street's reaction Nadella's memo landed two days after Anthropic CEO Dario Amodei published an essay calling for the industry to deliberately slow the pace of frontier model development. Amodei warned that increasingly capable AI agents could eventually cause serious economic disruption, cyberattacks, or worse. Sam Altman, Google DeepMind's Demis Hassabis, and Elon Musk all backed the call. Markets didn't shrug it off. Nvidia fell around 3% that Monday. Intel dropped more than 5% in early trading, as investors priced in what a slower AI buildout would mean for chip demand, according to Bloomberg. [Sam Altman and Dario Amodei Can't Agree on What AGI Even Means](https://startupfortune.com/sam-altman-and-dario-amodei-cant-agree-on-what-agi-even-means/) Bloomberg reports that Sam Altman now calls AGI a sloppy term while Dario Amodei has all but stopped using it, calling it powerful AI instead. Their timelines range from Anthropic's late 2026 to Altman's few thousand days, or roughly 2032, and neither shares a testable benchmark with the other. - [what does artificial general intelligence actually mean](https://startupfortune.com/sam-altman-and-dario-amodei-cant-agree-on-what-agi-even-means/) - [why AI leaders disagree on AGI definition](https://startupfortune.com/sam-altman-and-dario-amodei-cant-agree-on-what-agi-even-means/) Nadella isn't asking anyone to pause. He runs the company that just poured tens of billions into OpenAI's infrastructure and is racing to ship its own MAI models. His memo reads less like Amodei's essay and more like a compliance memo. It's one written by someone who has to answer to enterprise customers, regulators, and shareholders all at once. Where Anthropic frames the danger as existential, Nadella frames it as operational: fail at safety and you don't get to keep operating. ## What this means for the businesses buying it That distinction matters for the businesses actually buying this technology. A company running Copilot or Azure AI Foundry inside its own workflows doesn't need a philosophical position on superintelligence. It needs to know whether the vendor's model was independently tested and whether it can be shut off cleanly. And whether using it exposes the company to a compliance problem down the line. Nadella's memo is effectively telling Microsoft's own customers what he expects them to start asking their AI vendors, including Microsoft itself. That's the real shift here. Governments have been slow to write binding rules for frontier AI, leaning instead on voluntary commitments and industry codes. Microsoft's draft Code of Conduct, with its six-week comment window, looks like an attempt to get ahead of that gap: to have an answer ready before a regulator asks for one. Other large AI vendors may follow with their own published codes and third-party testing regimes. If they do, Nadella's operational-license framing could end up shaping enterprise AI compliance more than any single piece of legislation passed this year. For now, the draft rules are open for public comment until late October. Nadella hasn't said what happens to a Microsoft team that fails the standard he just set. **Also read:** [A Google DeepMind Safety Researcher Quit and Warned AI Could Kill Us All](https://startupfortune.com/a-google-deepmind-safety-researcher-quit-and-warned-ai-could-kill-us-all/) • [MiniMax's H3 Max Video Model Topped Sora and Veo, Then Its Price Quadrupled](https://startupfortune.com/minimaxs-h3-max-video-model-topped-sora-and-veo-then-its-price-quadrupled/) • [Bill Gates Pledges $1 Billion So AI Doesn't Leave the Poor Behind](https://startupfortune.com/bill-gates-pledges-1-billion-so-ai-doesnt-leave-the-poor-behind/) [Microsoft slashes AI transcription prices 72% with new MAI-Transcribe-2](https://startupfortune.com/microsoft-slashes-ai-transcription-prices-72-with-new-mai-transcribe-2/) Microsoft AI launched MAI-Transcribe-2 at $0.10 per audio hour, a 72% cut from its own model five months ago, and says it beats OpenAI, Google and ElevenLabs on both accuracy and speed. The model tops the FLEURS benchmark across 60 languages and transcribes an hour of audio in about ten seconds. - [microsoft ai transcription price drop 2026](https://startupfortune.com/microsoft-slashes-ai-transcription-prices-72-with-new-mai-transcribe-2/) - [best transcription service beats openai google accuracy](https://startupfortune.com/microsoft-slashes-ai-transcription-prices-72-with-new-mai-transcribe-2/) *This article is posted in [AI News](https://startupfortune.com/category/ai/), check it out for more related stories.* ## Join the discussion [Open in the community →](https://startupfortune.com/community/) Almost there. Sign in and your reply posts straight away.