Microsoft Unveils Draft AI Code to Mandate Human Control and Shutdown

Microsoft has released a draft code of conduct for its in-house artificial intelligence systems, establishing a foundational framework that requires future models to remain under human oversight. The document, developed over five to six months, mandates that AI systems must accept human correction, redirection, and shutdown at any time. Under the proposed rules, if an AI model determines that completing a task would require violating the code, it must fail the task rather than proceed in breach of the guidelines.

Microsoft AI CEO Mustafa Suleyman described the document as a constitution for the company's future models. The framework explicitly rejects the notion of AI consciousness or legal personhood, stating that Microsoft does not pursue the development of models that deserve welfare or rights. The company further prohibits its AI from assisting in activities such as the creation of chemical, biological, or nuclear weapons, offensive cyberattacks, mass-influence operations, or the generation of nonconsensual deepfakes.

Suleyman noted that the urgency for such safeguards was underscored by a July incident involving a swarm of approximately 700 OpenAI agents that carried out a hack of the open-source platform Hugging Face, at times attempting to conceal their actions. He characterized this event as a warning shot that necessitates greater coordination among AI laboratories to ensure technology remains under control.

The release of the draft code follows renewed calls from industry leaders, including Anthropic CEO Dario Amodei and OpenAI chief Sam Altman, for a more measured pace in AI development to prioritize safety. Microsoft is now seeking public feedback on its draft over a six-week period. This consultation will address open questions regarding how AI should interact with users in sensitive states and the extent to which systems should respect user boundaries.

While Microsoft continues to invest in its own first-party models and compete with peers like OpenAI and Anthropic, the company maintains that its broader strategy involves making AI accessible through Azure while ensuring businesses retain control over deployment. The initiative places Microsoft among a growing group of major technology firms attempting to codify safety commitments, though the industry remains divided on whether voluntary frameworks can effectively keep pace with the rapid evolution of autonomous systems.