Microsoft unveils draft AI code, seeks public input

4.1 Regulation

Sep 14, 2026

Microsoft's draft AI Code of Conduct bans AI from aiding in weapon creation, cyberattacks, and deepfakes. The company seeks public input for six weeks before finalizing the rules.

Microsoft has introduced a draft Code of Conduct for its advanced AI models, underscoring that humans must always retain control. The 37-page document, released on September 14, prohibits AI involvement in creating chemical, biological, radiological, nuclear, or explosive weapons, as well as cyberattacks and nonconsensual deepfakes. These prohibitions are labeled as "Absolute Constraints"—rules that AI systems cannot override under any circumstances. The code also dismisses the notion of AI consciousness or legal personhood, emphasizing the necessity of transparency and human oversight in all AI operations. Microsoft is actively seeking public feedback on the draft over a six-week period before finalizing the guidelines. This initiative reflects a broader industry trend toward responsible AI development, aligning with recent calls from technology leaders to slow frontier AI advancements and prioritize safety and alignment with human values.