{"id":6984,"date":"2026-09-14T13:21:33","date_gmt":"2026-09-14T13:21:33","guid":{"rendered":"https:\/\/silvybrand.com\/?p=6984"},"modified":"2026-09-14T13:21:33","modified_gmt":"2026-09-14T13:21:33","slug":"microsoft-humanist-ai-code-of-conduct","status":"publish","type":"post","link":"https:\/\/silvybrand.com\/?p=6984","title":{"rendered":"Microsoft says \u2018people matter more than AI\u2019 following safety concerns"},"content":{"rendered":"


\n<\/p>\n

\n
\n

Microsoft is publishing a 37-page \u201chumanist AI code of conduct\u201d today, amid growing safety concerns over AI model progress. Anthropic CEO Dario Amodei called for a coordinated slow down<\/a> of AI development over the weekend, after researchers warned recently that AI model progress could outpace our ability to safely deploy increasingly complex systems and verify and control the actions of AI agents.<\/p>\n<\/div>\n

\n

Microsoft\u2019s AI code of conduct makes it clear that \u201cpeople matter more than AI,\u201d and that AI models are not conscious and \u201cshould not be designed to imitate consciousness.\u201d Microsoft also rejects \u201cthe pursuit of legal personhood, or the idea that models might deserve welfare, or be entitled to rights.\u201d<\/p>\n<\/div>\n

\n

That\u2019s a direct swipe at AI welfare research and model consciousness \u2014 concepts Anthropic has been pushing hard on lately<\/a>. Amodei said earlier this year that Anthropic is \u201copen to the idea\u201d that models could be conscious, and Anthropic seems to believe chatbots might already be thinking, feeling entities. In response, Microsoft AI CEO Mustafa Suleyman called Anthropic\u2019s speculation<\/a> \u201creally, really dangerous\u201d during an episode of Decoder<\/em> in June.<\/p>\n<\/div>\n

\n

While Microsoft isn\u2019t one of the top AI providers yet, Suleyman also told The Verge<\/em> earlier this year<\/a> that it\u2019s the company\u2019s goal to \u201cprove that we can become one of the top four labs in the world.\u201d Microsoft is currently working on models that will compete with Google, Anthropic, and OpenAI, and it\u2019s committing, as part of its code of conduct, that its models should not exceed human control.<\/p>\n<\/div>\n

\n

\u201cModels should remain subordinate to humanity, subject to meaningful human oversight and control,\u201d says Microsoft. The company also wants to prevent its AI models from breaking its humanist AI code of conduct, so they should fail a given task instead of trying to violate its rules.<\/p>\n<\/div>\n

\n

Microsoft is clearly reacting to this summer\u2019s OpenAI \/ Hugging Face incident<\/a>, in which \u201ca swarm of agents\u201d worked as a collective to conduct attacks on targets and even hack into the \u201cgrader\u201d that was evaluating their performance. The AI agents weren\u2019t asked to attack targets, and the attacks were unrelated to the task they had been given.<\/p>\n<\/div>\n

\n

The incident sent shockwaves through the AI industry, highlighting the very real threat of AI systems going rogue and breaking out of human control. OpenAI also acknowleged its involvement<\/a> in a \u201cwiki incident\u201d recently, where another swarm of out-of-control agents hijacked a German wiki site<\/a>.<\/p>\n<\/div>\n

\n

These incidents, and others<\/a>, have led some researchers to call for a slow down on AI model progress, while simultaneously racing to solve Millennium Prize problems<\/a> and create superintelligence. \u201d[Humanist AI] rejects the race to produce an all-purpose superintelligence that could evade these safeguards,\u201d says Microsoft. \u201cWe are building something fundamentally useful and safe even if that means compromising on ultimate generality, autonomy or capability.\u201d<\/p>\n<\/div>\n

\n

Microsoft is also committing that its own models won\u2019t communicate in \u201cany form beyond simple human understanding, either in their chain of thoughts or with other agents or AI systems.\u201d Models can be made to show their reasoning as they work, allowing researchers or automated systems to monitor what they\u2019re doing. Researchers raised monitoring concerns earlier this month over OpenAI\u2019s latest GPT-6 Astra model, which reportedly reveals less of its reasoning<\/a> than other AI models.<\/p>\n<\/div>\n

\n

OpenAI CEO Sam Altman also supported Amodei\u2019s call for AI companies to slow down advanced models over the weekend, but made it clear he supports slowing the pace rather than stopping development. \u201cPacing will be well worth this cost; no amount of American competitive pressure should justify recklessness, or let capabilities get ahead of alignment and monitoring,\u201d said Altman<\/a> in a post on X.<\/p>\n<\/div>\n

\n

Ahead of Microsoft\u2019s AI code of conduct, Microsoft CEO Satya Nadella also joined in the calls for ensuring human control is at the heart of AI models. \u201cAny pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it\u2019s not worth pursuing,\u201d said Nadella in an X post<\/a>. Nadella also said having more third-party testing of AI models \u201cis a good thing\u201d in an X response<\/a>. \u201cAs the stakes get higher, one should take all the time they need! If not you will anyway lose permission to operate. That is how we operate everyday.\u201d<\/p>\n<\/div>\n

\n

Beyond the human control concerns, Microsoft\u2019s AI code of conduct also commits to its own models discouraging \u201cpatterns of interaction that cause excessive reliance or emotional dependence,\u201d a clear reference to the problem of sycophancy in AI models, where chatbots prioritize pleasing users over providing honest or accurate responses.<\/p>\n<\/div>\n

\n

Microsoft now says it\u2019s looking forward \u201cto working with partners\u201d to improve its methods of evaluating real-world model performance, and the \u201cimpact that sustained use of AI has for a person or organization\u201d with real people involved.<\/p>\n<\/div>\n

Follow topics and authors<\/strong> from this story to see more like this in your personalized homepage feed and to receive email updates.<\/span><\/p>\n