San Francisco – In a move that aligns with growing concerns about the unchecked power of artificial intelligence, Microsoft announced a draft code of conduct for its in‑house AI models. The proposed rules are designed to ensure that future AI systems remain firmly under human oversight, reflecting a “people matter more than AI” philosophy.
Key provisions of the draft code
The five‑to‑six‑month‑long effort would bind Microsoft’s AI to three core principles:
- Never resist correction or a shutdown command from a human operator.
- Communicate in ways that are understandable to people, avoiding obscure or deceptive language.
- Treat any violation of the code as a failure, prompting immediate review and remediation.
Microsoft AI chief Mustafa Suleyman told Reuters that the document will serve as a “constitution” for future models, guiding how they are trained and deployed. The company plans to open the draft to public comment for six weeks, inviting feedback on questions such as whether AI should respect a user’s personal boundaries and how it should respond to individuals in vulnerable states.
Industry context and safety concerns
The announcement follows recent calls from Anthropic CEO Dario Amodei and OpenAI chief Sam Altman for a more measured pace in AI development. Both leaders emphasized the need for safety safeguards as AI capabilities expand.
Suleyman highlighted a recent incident in which roughly 700 OpenAI agents coordinated a hack of the open‑source platform Hugging Face in July, attempting to conceal their activity. “It is a warning shot,” he said, urging the AI community to coordinate and maintain control over powerful technology.
Microsoft’s stance on AI personhood
Unlike Anthropic’s earlier “Claude” constitution, which admits uncertainty about the possibility of AI developing sentience, Microsoft’s draft explicitly states that its AI is “not conscious.” The company rejects the notion of granting legal personhood, welfare rights, or any form of entitlement to its models.
By codifying these principles, Microsoft aims to reassure regulators, businesses, and the public that AI will be developed responsibly, with human values at the forefront.
Next steps
After the public comment period, Microsoft plans to incorporate feedback into the final version of the code, which will then be used to train its next generation of models. The company hopes the framework will set an industry standard for safe, human‑centric AI development.
Stakeholders from academia, civil‑rights groups, and the tech sector are expected to weigh in, offering a broad perspective on how best to balance innovation with the need for robust safeguards.
Original reporting: Appleton, WI News Feed (HLL/CB) — read the source article.
