Moratorium on AI in weapons, with a human approving every kill
Puts a legal moratorium on AI in military weapons and warfare systems. No AI may approve a kill command or pick a target without a human who has real authority to decide.
A moratorium applies to all use of artificial intelligence in any weapons or warfare system until Congress lifts it by law.
No AI system may approve a kill command or select a target. A human with real knowledge and authority from the American people must approve each one.
The approving human must be a named, trained person in the chain of command, and that person's name must appear in the record of every strike.
Every use of AI in a targeting system must be logged and reported to the House and Senate Armed Services Committees within 30 days.
Enforcement and penalties
Enforced by the Defense Department Inspector General and the House and Senate Armed Services Committees, with a civil claim against the United States for families of anyone harmed by an unauthorized AI strike.
Criminal: a commander or official who authorizes an AI kill decision without a human approver faces up to 20 years in prison.
Civil: removal from command, forfeiture of pay for the period of the violation, and a fine of up to $5 million, adjusted for inflation.
Why
A machine has no knowledge of the people it targets and no authority from the American people. A person must answer for every kill. This item makes that rule part of the law.
Bills already in Congress
No bill in Congress does this yet. The text on this page is the proposal.
Start a petition
A petition shows how many people want this. Copy the text, start a petition, paste it in, and share the petition link.
I am your constituent, and I vote. Please support a law that bans AI from approving a kill command or choosing a target without a human who has real knowledge and authority.
Moratorium on AI in weapons, with a human approving every kill. Puts a legal moratorium on AI in military weapons and warfare systems. No AI may approve a kill command or pick a target without a human who has real authority to decide.
The law I am asking for would do the following:
A moratorium applies to all use of artificial intelligence in any weapons or warfare system until Congress lifts it by law.
No AI system may approve a kill command or select a target. A human with real knowledge and authority from the American people must approve each one.
The approving human must be a named, trained person in the chain of command, and that person's name must appear in the record of every strike.
Every use of AI in a targeting system must be logged and reported to the House and Senate Armed Services Committees within 30 days.
It must have teeth:
Enforced by the Defense Department Inspector General and the House and Senate Armed Services Committees, with a civil claim against the United States for families of anyone harmed by an unauthorized AI strike.
Criminal: a commander or official who authorizes an AI kill decision without a human approver faces up to 20 years in prison.
Civil: removal from command, forfeiture of pay for the period of the violation, and a fine of up to $5 million, adjusted for inflation.
Please tell me in writing whether you will introduce, cosponsor or vote for this, and if not, why not.