Microsoft’s AI Code of Conduct aims to curb AI behavior

Microsoft has joined a growing list of AI vendors pledging to control the behavior of its AI models. In a draft Humanist AI Code of Conduct, the company stated that "AI should not exceed human control," and that models should remain subordinate to humanity with meaningful human oversight and control. Microsoft emphasized the importance of developing systems with clear purposes, evaluated against real-world impact, and rejecting the pursuit of an all-purpose superintelligence that could evade these safeguards.

The company acknowledged the challenges in limiting AI development while maintaining safety, stating that "under- and over-caution represent failure modes with different types of consequences." Analysts generally agreed that Microsoft's goal is laudable, but noted the lack of specifics and verification mechanisms. Some experts raised concerns about contradictions within the document, such as Microsoft offering AI models through government clouds for defense and national security workloads.

While Microsoft provided some details about its evaluation framework and phased releases, independent board-level safety oversight, verifiable data points, and kill switches were not yet in place.