Authority
Who can decide, approve, stop or override? Decision Authority makes explicit which rights belong to humans, AI systems, agents, and governance bodies.
What authority means inside Decision Infrastructure.
Decision Authority is the explicit right of a human, system, or organisational body to recommend, decide, approve, execute, interrupt, override, or accept accountability for a consequential decision.
Who, or what, is permitted to recommend, decide, approve, execute, override and remain accountable?
This is the question to bring into board papers, executive discussions, AI governance reviews, assurance work and material decisions where authority may be unclear.
What weak authority looks like.
The visible structure
Policies, roles, workflows and controls may all exist. The organisation may appear governed on paper.
The decision gap
Decision participation changes, but authority is never deliberately redesigned. Several actors contribute. Nobody can clearly explain who actually decided or who can stop it from being executed.
A small statement can reveal a larger infrastructure problem.
“The human is in the loop.”
That statement is incomplete. Which human? At what point? With what evidence? Holding which decision right? A reviewer may be present without possessing meaningful authority.
Start with one real decision.
Test Authority, not the policy.
Take one consequential AI-enabled process and map: Recommend → Decide → Approve → Execute → Override → Account. Then ask whether those rights were deliberately designed or simply emerged from the workflow.
Relevant thinking and practical instrument.
Published foundation
Fixing the broken AI governance playbook · CIO
Five Questions That Audit Professionals Will Need to Answer in 2026 · ISACA
Next canonical paper
The Human-AI Authority Problem
In development for The Decision Layer.
Relevant instrument
Decision Rights Map
Map the six decision rights across people, systems and governance bodies, then identify where practical authority differs from formal authority.
Who, or what, is permitted to recommend, decide, approve, execute, override and remain accountable?
Use the question on one consequential decision. Then follow the evidence into the next part of the canon.