AIRUNAMOK

Microsoft introduced a new Humanist AI code of conduct built around the principle that people must remain in meaningful control of increasingly capable AI systems. The framework emphasizes the ability to correct and shut systems down, transparency about AI behavior, respect for human boundaries, and resistance to treating AI systems as independent moral or legal actors.

This week’s edition reflects a noticeable change in the AI-governance conversation. The question is no longer simply whether humans are “in the loop.” The question is whether humans still possess real authority. Across policy, healthcare, cybersecurity, identity, workplace systems, and autonomous agents, organizations increasingly need to prove that people can set boundaries, understand actions, restrict permissions, interrupt execution, verify outcomes, revoke authority, assign accountability, and shut systems down. The board-level question is becoming: As AI gains greater capability and autonomy, does the organization still have meaningful control?

Human control is becoming a design requirement rather than a general AI principle. Boards should ask whether advanced AI systems can be meaningfully interrupted, corrected, overridden, audited, and shut down — and whether those controls remain effective as systems become more autonomous.