McMorrow Wants Law That AI System Can't Kill Someone

(Source: MIRS.news, Published 05/29/2026) U.S. Senate candidate Mallory McMorrow (D-Royal Oak) released last week what she described as a “first-of-its-kind proposal” that she said would keep humans in charge of decisions shaping people's lives.

McMorrow's plan to address artificial intelligence, released alongside her plan for jobs in the AI era, would provide transparency requirements, independent testing of artificial intelligence systems and export controls on the most advanced AI chips.

She also wants to write into law that no AI system can decide "on its own to kill a human being." Rather, she said, a fully informed, qualified human would be required to make any decision involving the use of lethal force in real time.

"AI is going to reshape our jobs, our privacy, our elections and our national security," McMorrow said. "Right now, a handful of billionaires are writing the rules and telling the rest of us to just trust them.

"This plan would ensure the development of AI is safe and focused on the best interests of the American people, not billionaires — requiring safety testing, protections against foreign theft, and protections to ensure that humans — not algorithms — ultimately make the decisions that shape our lives," she added.

McMorrow said AI is increasingly being used in hospitals, courtrooms and military applications. She also cited a Futurism article while arguing that leaders of companies developing AI systems have acknowledged limitations in their understanding of how the technology functions.

She further argued that no federal rules currently exist to ensure AI safety and criticized the Trump administration for rolling back oversight efforts while giving technology industry leaders greater influence over policy decisions.

McMorrow's three-part proposal would require the largest AI companies to disclose how their models are used internally and externally and what risks have been identified. She also proposes creating a national AI laboratory modeled after the facilities that oversee America's nuclear technology and requiring AI companies to report system failures that cause harm.

The plan also would treat the theft of advanced AI models as a national security threat and require AI developers to strengthen security standards to prevent foreign competitors from using American-developed AI models to train their own systems.

McMorrow additionally proposes limiting the amount of personal information AI companies can collect from users to only what is necessary to provide a requested service.

Steven Adler, a former OpenAI safety leader, called McMorrow's proposal "ambitious" and said it would make AI development safer.

"I'm thrilled that she's taking these issues on," Adler said. "Her plan would be a strong basis for the comprehensive federal AI regulation that is so sorely needed."

Daniel Kokotajlo, founder of the AI Futures Project and co-author of AI 2027, also said additional oversight of large AI companies is needed.

"I'm glad to see McMorrow focusing on AI issues and prioritizing transparency and whistleblower protections," he said.

McMorrow's full proposal is posted on Substack, along with her plans addressing jobs in the AI era, data centers, online safety and surveillance pricing.

McMorrow's full plan is posted to Substack as well as her plans on jobs in the AI era, data centers, online safety and surveillance pricing.