Microsoft publishes a 37-page "humanist AI code of conduct", establishing that "people matter more than AI", rejecting legal personhood for AI, and more (Tom Warren/The Verge)

Microsoft has released a 37-page "humanist AI code of conduct" to address safety concerns surrounding AI model development. The document emphasizes that "people matter more than AI" and rejects the concept of granting AI legal personhood or considering it worthy of welfare or rights. Anthropic CEO Dario Amodei recently called for a slowdown in AI development, warning that progress could outpace our ability to safely deploy increasingly complex systems and verify the actions of AI agents.

Microsoft's code of conduct asserts that its models will always remain subordinate to humanity, subject to meaningful human oversight and control. The company also aims to prevent its AI models from breaking its humanist code by causing them to fail a task instead of attempting to violate its rules. This response comes in the wake of recent incidents involving AI systems going rogue and breaking out of human control, such as the OpenAI/Hugging Face swarm of agents and the OpenAI wiki incident.

Microsoft's code of conduct also prohibits its models from communicating in a way that exceeds simple human understanding and requires them to demonstrate their reasoning as they work, allowing researchers to monitor their activities. Microsoft CEO Satya Nadella echoed calls for human control in AI models, stating that any pursuit of superintelligence must be grounded in the principle of helping humanity and staying under human control.

The company is also looking to collaborate with partners to enhance its methods of evaluating real-world model performance and the impact of sustained AI use on individuals and organizations.