Minimum Privilege

Minimum privilege is a security principle requiring users and systems to operate with only the lowest level of access permissions necessary to perform their specific functions. In AI governance, this principle becomes critical for controlling model access, data handling, and algorithmic decision-making across organizations. Enforcing minimum privilege reduces the risk of unauthorized model manipulation, limits exposure to sensitive training data, and helps organizations meet compliance requirements like GDPR and regulatory standards for AI systems.

1 item

Corporate PolicyGlobal2026-09-15

Microsoft's MAI Code of Conduct Sets Agentic Chain-of-Command Standard

Microsoft published a draft 'Humanist AI Code of Conduct' for its MAI Models on September 15, 2026. The policy introduces Absolute Constraints blocking offensive cyber outputs and a Chain of Command authority model for agentic systems requiring minimum-privilege operation. A six-week public consultation precedes a revised version intended to govern 2027 model development.

Microsoft MAI Models agentic AI dual-use AI offensive AI constraints chain of command minimum privilege model conduct standards