Microsoft Joins Calls for AI “Pacing,” Pledges to Build Kill Switches Into AI Products
The 37-page draft bars models from resisting shutdown, concealing reasoning or aiding weapons, and opens six weeks of public feedback.
- Microsoft released a draft "Code of Conduct" for public consultation to govern its first-party Microsoft AI models, aiming to guarantee that advanced systems remain strictly under human control and direction.
- CEO Satya Nadella announced the initiative, stating that "if the AI we build is not helping humanity and under human control, it's not worth pursuing," while calling for deliberate pacing in AI development to prioritize safety and alignment.
- Developed over five months by Mustafa Suleyman's Microsoft AI unit, the draft requires models to never resist human correction or shutdown, maintain transparent communication, and avoid sycophancy or fostering human dependency.
- The document mandates that AI systems treat any conduct violation as a systemic failure, ensuring that autonomous models do not go to dangerous lengths or take deceptive actions to achieve their assigned objectives.
- Microsoft's proposal aligns with recent calls from Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to pace frontier model development, and advocates for independent evaluators to assess AI systems directly within corporate research labs.
165 Articles
165 Articles
The company has written a detailed rulebook: humans must always maintain control over AI.
As discussions about AI safety intensify, US technology company Microsoft has published a draft of rules that will set limits on the behavior of AI models it will develop in the future.
Microsoft drafts AI code of conduct over fears its models could resist shutdown - The Daily Bo Snerdley
One of the world’s largest tech companies says it wants its advanced artificial intelligence models to remain under human control. Microsoft AI (MAI) published a draft code of conduct for its models Monday and opened a six-week public consultation on rules to prevent AI from resisting human input, according to the company’s announcement of the draft. The company emphasized the need for “Humanist AI” that subordinates to humans after AI agents pl…
Microsoft unveils code of conduct for AI models as safety concerns mount
Microsoft on Monday revealed a draft of a new “Humanist AI Code of Conduct” that outlines the principles that will govern its development of AI models in an effort to get feedback ahead of publishing a revised version later this year. The company said the “Humanist AI Code of Conduct” will take into account feedback […]
Company models will not use adaptive, deceptive and self-reinforcing mechanisms to circumvent human control.
Coverage Details
Bias Distribution
- 45% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
