Weekly newsletter
Every Tuesday, a short field note on what was approved, who owns the decision, and what evidence survives after deployment.
June 23, 2026
Three major AI governance announcements landed in the first week of June 2026: Executive Order 14409, Microsoft's Agent Control Specification, and Microsoft's updated AI Red Team taxonomy. All three address what agents do while running. None answers who authorized the agent's scope before it ran the first task.
Read issue 15 →June 16, 2026
An internal auditor asks who approved the ten thousand dollar auto-approval threshold inside an accounts payable agent, and no one in the room has the record. Microsoft's Agent Governance Toolkit, released April 2, 2026, can prove the rule was enforced. It cannot prove the rule was ever reviewed under the accountability model the AI agent was supposed to operate under.
Read issue 14 →June 9, 2026
At Build, Microsoft introduced agents that act without being prompted and a tuning loop that lets them keep changing after you authorize them. Your governance dashboard reads green while your authorization quietly goes stale. This week is about the gap the coverage ratio cannot see, and the one field that closes it.
Read issue 13 →