US House bill to ban advanced 'superintelligence' AI development introduced

Reporting by the420.in describes a renewed push in the United States for tighter controls on frontier AI after Democratic gains in the midterm elections. On November 4, Congresswoman Yassamin Ansari argued the Democratic agenda should include pausing advanced AI, regulating the industry more firmly and banning artificial superintelligence. Her remarks renewed interest in bills introduced earlier this year. No ban has been enacted; these are bills awaiting action.

The House measure is H.R. 10538, the Ban Artificial Superintelligence Act of 2026, brought forward by Representative Greg Casar on September 24, with Ansari among the co-sponsors. Senator Bernie Sanders introduced a companion bill, S. 5493, on September 23. The legislation would set up a federal Department of Artificial Intelligence and require a pause on systems classed as advanced AI until that department is established and safety rules exist. Advanced AI is defined mainly by training compute, at a threshold of 10^25 operations or more that could shift over time.

While the pause lasts, training, modification and fine-tuning of these systems would be off limits, except in narrow cases such as stripping out precursor capabilities. Unreleased advanced systems could not be deployed until rules on monitoring, audits and approval exist, and developers would eventually need a federal charter. Superintelligence itself could not be developed, funded, held, imported or transferred. Reckless violations by covered policymakers or rogue actors could bring fines and up to 20 years in prison, while other individuals could be barred from the AI industry for 10 years.

Both bills are in committee. The House version was sent to the Oversight and Government Reform panel, and the Senate version to the Commerce, Science and Transportation panel. Neither has cleared its chamber.