Lethal Autonomous Weapons Systems (LAWS) — "slaughterbots" or "killer robots" — can select and engage targets without human intervention. The technology largely exists; the question is whether we deploy it.
The core objection
Delegating the decision to kill to an algorithm crosses a moral line. A machine has no judgment, no mercy, and no one to hold accountable when it's wrong.
Why it's destabilizing
| Risk | Why |
|---|---|
| Accountability gap | If an autonomous weapon commits a war crime, who's guilty? |
| Lower threshold for war | No soldiers at risk makes conflict politically cheaper |
| Speed | Machine-vs-machine escalation faster than humans can intervene |
| Proliferation | Cheap drones + open models = accessible to many actors |
The debate
A global "Campaign to Stop Killer Robots" pushes for a treaty banning fully autonomous lethal weapons, keeping meaningful human control over force. Major militaries resist binding limits, citing strategic necessity — a textbook coordination failure.
It's the control problem with live ammunition, and the clearest near-term catastrophic-risk application of AI.
Related: The Control Problem · Moloch and Coordination Failures · AI Regulation