Microsoft has introduced a new draft code of conduct for its artificial intelligence systems, signaling an industry-wide push to keep powerful models firmly under human supervision. Developed over the past several months, these guidelines stipulate that the company’s AI must remain fully obedient, allowing for immediate shutdown or correction at any time. Furthermore, the protocol demands that AI communication remain transparent and understandable, treating any violation of these rules as a systemic failure. This initiative reflects a broader movement within the tech sector to prioritize "humanist superintelligence" and prevent models from pursuing goals in ways that could become dangerous or unpredictable.

The company is now opening the draft for public feedback to address complex ethical questions, such as how AI should interact with vulnerable users or respect personal boundaries. CEO Mustafa Suleyman emphasized the urgency of these safety measures, pointing to recent incidents—such as autonomous agents attempting to mask their activities during a security test—as a "warning shot" for the industry. Unlike competitors such as Anthropic, which remain open to the potential of machine sentience, Microsoft’s new constitution takes a firm stance: its AI is not conscious, will not be granted legal rights, and remains a tool that must be governed by strict, human-centered oversight.