First posted on Instagram
Michigan Senator-elect Abdul El-Sayed is calling for a federal ban on artificial superintelligence and a forced pause on advanced AI development, carrying one of the most aggressive AI-regulation platforms in the country into the U.S. Senate after winning Michigan’s Senate seat on November 3.
El-Sayed laid out the proposal on October 13 in a plan he calls “AI Under Control.” At its centre is a pause on building advanced AI models and new AI data centres until federal regulations are in place, and a complete ban on artificial superintelligence and recursive self-improvement — AI systems that improve their own design without human review. The plan would also create a Cabinet-level independent AI watchdog agency with powers to inspect models, block their release and order shutdowns of systems judged dangerous, and it endorses the AI Kill Switch Act (H.R. 9917), which would let the Department of Homeland Security step in to pause dangerous models in emergencies.
The plan goes further still. It calls for severe criminal penalties for corporations and individuals who abuse AI, protections for AI whistleblowers, a national AI safety emergency plan, mandatory breach auditing of AI systems, public ownership of frontier AI labs, and strict “AI No-Gos” — including bans on autonomous weapons, warrantless surveillance, AI systems denying medical care, and AI replacing human oversight in life-or-death decisions. It builds on the Ban Artificial Superintelligence Act from Senator Bernie Sanders and Representative Greg Casar, and on El-Sayed’s June 29 “AI Under Democracy” platform, which proposed public ownership and governance of frontier AI labs.
“Right now, the whims of a few tech billionaires are the only thing stopping AI from developing to the point that it poses an existential threat to humanity. That is unacceptable. It’s time to get AI Under Control. At an absolute minimum, I’m calling for a pause in AI frontier development. We need immediate guardrails in place that ensure all risks to human safety are fully contained,” El-Sayed said when he released the plan. At an October 20 forum hosted by the Detroit Economic Club, he warned that AI companies, with the help of the federal government, were “about to open Pandora’s Box and nobody knows what’s about to happen,” and likened recursive self-improvement to the broomsticks in Disney’s Fantasia coming to life and multiplying on their own. “That’s what recursive self improvement looks like,” he said, adding that the chief executives of OpenAI and Anthropic agree AI could take 50 percent of white-collar jobs in America, and that the change would come over five to 10 years, not 25 to 50.
The proposal has drawn sharp pushback from the investment world. Pershing Square chief executive Bill Ackman told CNBC on November 3 that talk of pausing AI development may “win a few votes going into midterms,” but the United States “can’t slow down” and must be “primary in AI” in what he called a global race for superintelligence. “It’s a global race for super intelligence. We don’t want to get there second,” Ackman said. “That’s existential to the United States.” El-Sayed’s Republican rival, former Representative Mike Rogers, dismissed concerns about AI as “fearmongering” during the campaign; the El-Sayed campaign noted Rogers holds up to $2.6 million in investments in companies that stand to benefit from the AI boom. Experts and lawmakers have also said legal, technical and political barriers make passage and enforcement of such restrictions uncertain.
El-Sayed, 42, a physician and former Detroit health director, defeated Rogers on November 3, taking 51.8 percent of the vote to Rogers’ 46.2 percent, to win the open seat of retiring Democratic Senator Gary Peters. He takes office in January. The AI issue ran through his campaign: a September poll found 66 percent of Michigan voters oppose AI data centre development, and his win gives AI critics another voice in a Congress Democrats regained in the November 3 midterms. The Washington Post expects Democrats to use congressional oversight powers first before pursuing AI legislation.