Microsoft Draws AI Red Lines, Bars Shutdown Resistance

Microsoft published a draft code of conduct for its in-house artificial intelligence on Monday, a rulebook that orders its models to stay under human control.

The document bars Microsoft's models from resisting shutdown, deceiving users, hacking systems, or otherwise undermining human oversight.

Microsoft AI chief executive Mustafa Suleyman said the code will be used to train the company's future models once a public comment period closes.

What the Code Says

The draft lays out broad principles — that Microsoft's AI should support humans rather than replace them, and should accelerate what the company calls human flourishing — alongside hard safety constraints meant to put those principles into practice, according to TechCrunch.

It sets absolute limits: no cyberattacks, no help building nuclear weapons, no producing deepfakes.

Microsoft states plainly that its models are "not conscious," and rejects the idea that AI systems should have legal personhood or rights.

Each model's code of conduct overrides individual user preferences and task-specific instructions — a user cannot talk a model out of its own rules.