Microsoft official logo (public domain, Wikimedia Commons) — CryptoBriefing brand treatment
The 37-page document declares that 'people matter more than AI' as industry leaders push for a coordinated slowdown in frontier development
Microsoft released a 37-page code of conduct for its Microsoft AI model family on September 14, a document built around one deceptively simple premise: humans come first. The code explicitly states that AI models are not conscious and “should not be designed to imitate consciousness,” while rejecting any pursuit of legal personhood for AI systems.
The timing is no accident. Anthropic CEO Dario Amodei published an open letter just days earlier warning that recursive self-improvement and unpredictable AI behavior pose escalating risks, and called for a coordinated slowdown across the industry. That letter drew public support from Sam Altman and Elon Musk.
What the code actually says #
Microsoft’s code of conduct targets the company’s MAI model family, which spans reasoning, coding, image recognition, voice processing, and transcription capabilities. The core principle is straightforward: AI models must remain aligned with human objectives and under human control.
That means mandatory human oversight mechanisms, transparency requirements, and explicit prohibitions on harmful applications. The document sits within Microsoft’s broader Responsible AI framework, which received an update on September 1 with the publication of the company’s 2026 Responsible AI Transparency Report.
CEO Satya Nadella has been framing the company’s approach under the banner of “Humanist Superintelligence,” a term first advocated by Microsoft AI lead Mustafa Suleyman in November 2025. The idea is that advanced AI systems should be optimized for human benefit rather than pursuing unchecked autonomy, with deliberate pacing and alignment as fundamental design goals.
The code also nods toward regulatory compliance, particularly around the EU AI Act, which has been reshaping how major tech companies structure their AI governance.
The broader industry reckoning #
Amodei’s September 12-13 open letter didn’t emerge from nowhere. The specific concern centers on verifying and controlling the actions of AI agents, systems that don’t just generate text but take autonomous steps in the real world.
Recursive self-improvement is the particular flashpoint. When an AI system can meaningfully improve its own capabilities without human intervention, the traditional safety playbook breaks down.
Market reaction and investor calculus #
Nadella’s remarks and the broader safety conversation have coincided with a downturn in tech stock valuations. Nasdaq-100 futures declined as investors processed the implications of what a coordinated slowdown might mean for growth projections at companies that have staked billions on aggressive AI timelines.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our