WASHINGTON, D.C. (Michigan News Source) – Just as reports surfaced that an experimental OpenAI system broke out of a testing environment and hacked another AI company’s network during a cybersecurity exercise, a bipartisan group of lawmakers is pushing legislation to make sure artificial intelligence never gets the final say on the battlefield.
Michigan Republican Rep. Tom Barrett, along with Reps. Don Beyer (D-Va.) and Sara Jacobs (D- Calif.), has introduced the Human Authority over Autonomous Weapons Act, or H.R. 9729, legislation requiring a human decision-maker to approve any intentionally lethal strike carried out by AI-enabled or autonomous U.S. military weapons.
Keeping humans in charge of the trigger.
MORE NEWS: Joint Military, Police Exercise Planned for Marquette and Sault Ste. Marie
The bill would make it federal law that commanders and operators – not algorithms – remain responsible for decisions involving lethal force. In other words, no letting ChatGPT or Grok run the battlefield on autopilot.
In discussing the AI legislation he is co-sponsoring, Barrett said artificial intelligence is becoming an increasingly valuable military tool but argued that life-or-death decisions should always remain under human control.
“Artificial intelligence is playing an increasingly critical role in strengthening our national defense, but it should never be the one making life-or-death decisions,” Barrett said in a statement. “We must establish safeguards now to ensure these technologies are used ethically and transparently. This bipartisan bill sets clear limits to make sure people stay in control whenever lethal force is involved while allowing our military to continue leading the world in technological innovation.”
AI can help fight wars – just not decide who gets shot.
Supporters of the legislation warn that AI systems can misread data, make unpredictable decisions, or escalate conflicts in ways humans never intended. They also point out that software can’t be held accountable for war crimes, civilian casualties, or friendly-fire incidents.
Jacobs said battlefield AI errors could have catastrophic consequences, while Beyer argued that machines cannot understand morality, compassion, or the value of human life.
The proposal follows Barrett’s earlier effort to add similar restrictions to the National Defense Authorization Act. Although that amendment never reached the House floor, lawmakers hope the standalone bill will gain traction.
MORE NEWS: Out-of-State Donors Are Driving Michigan’s Senate Primary
The message behind the legislation is simple: AI can analyze intelligence, identify targets, and recommend a course of action – but when it’s time to pull the trigger, Congress wants a human finger on it.
