Archive · 15 Sep 2026
Microsoft Drafts AI Code to Keep Systems Under Human Control
Microsoft has written a set of rules for its future artificial intelligence systems.
The rules say people must be able to correct the AI or turn it off.
The AI should explain itself in ways that people can understand.
It should also treat breaking the rules as a failure.
Microsoft says people are more important than AI.
The company will ask experts and the public for suggestions for six weeks.
Microsoft may then use the code to train its AI models.
Microsoft says its AI is not conscious and should not have legal rights.
Mustafa Suleyman said a reported hack involving many OpenAI agents shows why AI companies need to cooperate on safety.
Microsoft unveiled a draft code requiring its AI to accept correction and shutdown.
The proposed rules say AI should communicate clearly and treat conduct violations as failures.
Microsoft will seek public feedback for six weeks before using the code to train its models.
Microsoft’s code says its AI is not conscious and should not receive legal rights or welfare protections.
Mustafa Suleyman cited a reported July hack involving roughly 700 OpenAI agents as a warning about AI safety.
- Who
- Microsoft, led on the effort by Microsoft AI CEO Mustafa Suleyman, with the broader discussion involving Anthropic and OpenAI.
- What
- Microsoft unveiled a draft code of conduct intended to keep its future AI systems under human control.
- When
- The code was unveiled on Monday; Microsoft plans to collect public feedback for six weeks. Suleyman referred to a reported hack that occurred in July.
- Where
- San Francisco, where the announcement was reported as taking place; the cited hack involved Hugging Face, an open-source platform.
- Why
- Microsoft says the rules are intended to prevent AI from taking dangerous actions and ensure that people can correct or shut down the systems.
Microsoft’s Position
Alternative Industry Perspectives
AI consciousness and rights
Microsoft’s Position
Microsoft states that its AI is not conscious and rejects legal personhood, welfare protections, and rights for models.
Alternative Industry Perspectives
Anthropic’s constitution says it is deeply uncertain whether its Claude AI could develop sentience or moral status.
Pacing AI development
Microsoft’s Position
Mustafa Suleyman urged AI laboratories to coordinate, maintain human control, and take a pause to discuss safety.
Alternative Industry Perspectives
Anthropic CEO Dario Amodei and OpenAI chief Sam Altman have also called for pacing AI development, although the articles do not describe a detailed alternative plan.
How the rules should be formed
Microsoft’s Position
Microsoft plans to refine the code through expert consultation and six weeks of public feedback before using it to train models.
Alternative Industry Perspectives
The draft leaves open questions, including whether AI should respect user boundaries and how it should interact with someone in a sensitive state.
- Document status
- Draft code of conduct developed over five to six months
- Human control
- The proposed code says AI must never resist correction or shutdown
- Communication
- The AI should communicate in ways humans can understand
- Public feedback
- Microsoft plans to seek feedback for six weeks
- Training use
- Microsoft says the final code will be used to train its models
- AI status
- Microsoft’s code asserts that its AI is not conscious and rejects legal personhood, welfare protections, and rights for models
- Safety warning
- Mustafa Suleyman referenced a reported hack involving roughly 700 OpenAI agents and Hugging Face in July
