Human-in-the-loop isn't human authority: scoped grants beat click-approval fatigue

arthaudm · reddit · 2026-09-13

"Require human approval" sounds safe until every action needs a click: people stop reading and approval becomes muscle memory—the human is technically in the loop and practically absent. The author proposes scoped authority instead: each grant specifies who granted it, the exact action and target, limits and expiry, what evidence the agent must use, and what must return to a human. Routine actions run inside the boundary; anything outside stops. The tool enforces the grant rather than trusting the prompt to remember policy. The real design question isn't whether a human touches every action, but whether the agent can prove which human had authority for it.

Original post →

More from coding & agent

coding & agent channel →