Microsoft Opens Its AI Rulebook to Public Review
3 min readMicrosoft is publishing the behavior rules behind its own AI models and inviting the public to argue with them. Chairman and CEO Satya Nadella said on September 13 that the company would release a Code of Conduct governing its first-party MAI models the following day, and open it for public consultation rather than keep it as an internal document.
What Prompted the Move
The announcement followed an essay by Anthropic CEO Dario Amodei titled “We Must Pace the Frontier,” published over the weekend. Amodei argued that AI capabilities are now improving faster than researchers can understand or control them. He proposed independent evaluators working inside the major labs, shared safety standards across democratic countries, and eventual international limits on the most dangerous capabilities such as recursive self improvement.
Model behavior policies are not new. OpenAI, Anthropic and Google each publish some version of a specification describing how their systems should respond and where the limits sit. The difference here is the consultation step. Microsoft is putting its rulebook out for comment before treating it as settled, which invites outside critics to shape a document that would normally be written entirely in house.
Nadella Draws a Line on Superintelligence
In his post, Nadella wrote that any pursuit of superintelligence has to be grounded in the principle that AI which is not helping humanity and under human control is not worth pursuing. He said Microsoft welcomes the deliberate pacing needed to make alignment a design goal rather than an afterthought, and singled out the idea of embedded evaluators as worth building real mechanisms around.
He was not alone. Sam Altman said OpenAI would give independent evaluators employee level access to its systems, and Elon Musk and Google DeepMind chief Demis Hassabis both voiced support for Amodei’s framing. For a field that has spent three years treating release cadence as the main scoreboard, a week of rivals publicly agreeing to ease off is unusual.
Why the Microsoft AI Code of Conduct Matters
Agreement among lab leaders is the easy part. Wall Street and the Trump administration have both built expectations around AI moving fast, and the first market reaction was blunt: SoftBank fell more than 11 percent in Tokyo on Monday after OpenAI ruled out a 2026 listing and backed a slower pace.
The consultation is the part worth watching. A code that absorbs outside comment and turns into auditable commitments would mark a real change in how these rules get written. One that closes quietly with the text unchanged would confirm every skeptic who called the safety turn a press cycle.
Microsoft’s rulebook goes live for public comment on September 14. Whether the other labs follow with their own open drafts is the next signal to watch.