Glossary · 6 · Quality, risk and governance
Agent kill switch
Also known as: Kill switch (AI agent), Agent stop, Agent halt
German: Not-Aus für Agenten (umgangssprachlich)
An agent kill switch is a control outside the AI agent that stops one agent, a group of agents or all agents of a platform at once — ending running tasks, blocking new tool calls and revoking or suspending their access — without needing the agent’s cooperation.
- Intermediate
- Technical project managers
- Technical writers
- Developers
In one sentence
An agent kill switch stops AI agents from outside: tasks end, tool calls are blocked, access is suspended. It is not a machine emergency stop.
Example
When an agent starts mass-mailing customers, the operator presses “Stop all agents” in the agent platform: the queue is frozen, the mail tool is blocked and the agent’s token is suspended within seconds.
Explained in context
This term is a station of a mini course
- When an AI agent crosses the line: detect, contain and restore operationsStation: The containment ladderStart the course
Context cards connect this term with others to answer one question. Also in British English and German.
- Containing an AI agent: pause, revoke, isolate, shut downHow do you stop an AI agent that has crossed its boundary — without making things worse, and without confusing it with stopping a machine?UKDeutsch
- When an AI agent crosses the line: detect, contain and restore operationsWhat has to be in place when an AI agent does something it isn’t allowed to do — in IT systems, IoT fleets, OT and on machines — and how is it detected, stopped and released again?UKDeutsch
How it applies
- Outside the agent: A switch the agent can ignore, argue with or delay is no switch. It acts on the platform, the gateway and the credentials, not on the conversation.
- Levels: One task, one agent, one agent type, all agents. The levels are the first rungs of agent containment.
- Tested like any stop: Whether the switch works is known only after it has been tried — including for agents that run in the background or call other agents.
- The AI Act's stop: For high-risk AI systems, Article 14(4)(e) requires that people can interrupt the system through a “stop” button or a similar procedure that brings it to a halt in a safe state — see stop button or similar procedure.
Agent kill switch vs. emergency stop
The name borrows from machinery, but the two are different things. An emergency stop is a safety function of a machine, designed and validated under machinery law and standards such as ISO 13850, and it stops hazardous motion. An agent kill switch stops software from deciding and calling tools. Where an agent controls a machine, an agent stop must never be presented or wired as the emergency stop, and the machine's own stopping functions and safe state stay with its control system.
Why it matters on your learning path
- Technical writers: Name it precisely — “stop agents”, not “emergency stop” — in interfaces and instructions.
- Technical marketers: Don't sell a kill switch as a safety feature of a machine.
- Technical project managers: Name who may press it and rehearse it before go-live.
- Developers: Implement it at the gateway and token level, with an audit record of every use.