Skip to main content
Delegrity

Blog

Notes on authority for AI agents.

Plain notes on delegating real actions to AI agents: authority, approval and evidence.

  1. When two AI agents are both authorised, who decides?

    Single-agent controls ask whether an agent may act. Systems of agents add a harder question: what happens when two legitimately authorised agents collide.

    5 min read

  2. Access is not authority: why valid tokens still allow wrong actions

    An agent can pass every identity and token check and still take an action nobody authorised. Here is why, and what a per-action authority check adds.

    5 min read

  3. What the 2026 data says about AI agents acting without authority

    Surveys, incidents and new guidance point to one gap: agents with valid access taking actions nobody authorised, and organisations unable to show what happened.

    4 min read

  4. What changes after approval? The beneficiary-swap problem in AI agents

    An approval is only as good as its link to the action that runs. Why approving a request type is not enough, and how a recheck before execution closes the gap.

    5 min read

  5. Mandates, not permissions: how organisations should delegate to AI agents

    Permissions say what an agent can reach. A mandate says what it may do, for whom, within which limits, and until when. Why delegation to agents needs the second.

    5 min read