Skip to main content

Weekly newsletter

AI accountability after the demo is over

Every Tuesday, a short field note on what was approved, who owns the decision, and what evidence survives after deployment.

The Governance Gap
Reset

August 11, 2026

The Governance GapGovernanceAgents
Latest

The Agent Was Authorized. The Schedule Never Was.

Routines in Foundry Agent Service let an agent run automatically on a timer, a schedule, or an external event. The routine record holds the trigger, the action, and the run history, and none of the three records who decided the run should exist or when anyone will ask whether it still should.

Read issue 22 →

August 4, 2026

The Governance GapGovernanceAgents

Your Chatbot Became a Compliance Test on Saturday.

The Digital Omnibus deferred the AI Act's high-risk chapter to December 2027, but the Article 50 transparency obligations applied on schedule from August 2, 2026. Any customer, competitor, or national authority can test whether your chatbot discloses what it is, which is why enforcement starts with the checks that cost nothing to run.

Read issue 21 →

July 28, 2026

The Governance GapGovernanceAgents

Microsoft Automated Your Agent Count. Nobody Automated Your Coverage.

Microsoft's Agent 365 registry, generally available May 1, 2026, answers the question governance committees spent two years on: how many agents do we have. It does not answer how many of them carry an authorization record, a named owner, and a review date that has not lapsed, and no vendor release is coming for that number.

Read issue 20 →