Contraction trigger

What Is a Contraction Trigger for AI Autonomy?

In short

A contraction trigger is a condition, written down in advance, that narrows what an AI agent is allowed to do. Examples are a model change on a consequential path, a new scope, a rising exception rate, a near miss, or a regulatory change. Writing them down beforehand is what makes them usable under pressure.

Updated 24 September 20261 min read

Why does autonomy need to contract as well as expand?#

An operating model that can only widen autonomy is a ratchet, and a ratchet is the wrong mechanism for a control that governs risk. Trust earned under one set of conditions does not survive a change to those conditions, and a model swap or a new class of work can invalidate months of evidence in a day.

Keep up with the Knowledge BaseEvery two weeks, get new terms and updated definitions straight to your inbox.

What makes a trigger work?#

Being written down before it is needed. A contraction decided in the middle of an incident is an argument; one agreed in advance is a procedure. It names the condition, what narrows, and who is told.

Triggers are the counterweight to automation follows trust. Together they make the delegation boundary a line that moves in both directions for stated reasons, rather than one that only ever loosens.

Found something wrong, missing or badly explained? Tell us. We credit every correction we publish.

Why does autonomy need to contract as well as expand?#

What makes a trigger work?#

Related terms