Sen. Bernie Sanders (I-VT) and Rep. Greg Casar (D-TX) introduced legislation Thursday that would impose criminal penalties, including up to 20 years in prison, on developers of advanced artificial intelligence systems deemed to exceed certain safety thresholds.
The bill, called the Ban Artificial Superintelligence Act, would temporarily halt development of the most advanced AI models and create a new Cabinet-level federal agency empowered to order the physical destruction of noncompliant systems. The legislation compares the development of highly capable AI to the unlawful production of nuclear weapons, and includes provisions critics have described as a “corporate death penalty” for violating firms.
Specifically, the measure targets AI systems capable of matching or exceeding human cognitive performance, and would outlaw any model shown to be capable of subverting shutdown commands, carrying out unauthorized cyberattacks, or undermining the federal government.
Sanders and Casar have framed the bill as a necessary check on what they describe as unaccountable “AI oligarchs” building systems with no meaningful oversight. Neither lawmaker’s office released additional comment beyond the bill’s introduction as of publication.
Experts Warn of Competitive Fallout
Technology policy analysts have raised concerns that the bill, if enacted, would slow U.S. AI development without doing anything to constrain rivals such as China and Russia, who are not bound by American law.
“If the United States stops developing advanced AI, China is unlikely to follow. Surrendering the lead in a potentially transformative technology would carry serious economic and national security costs,” said Daniel Castro, president of the Information Technology and Innovation Foundation (ITIF), a nonpartisan tech policy think tank.
Gary Marcus, an AI researcher who has testified before the U.S. Senate on the subject, offered a more mixed assessment. He said a “permanent, unilateral ban on all research into superhuman AI is too broad” and warned the bill amounted to “a guarantee” that the U.S. would fall behind in AI development. Marcus called the proposal “not the right approach,” though he acknowledged some elements of the bill were “defensible.”
Long Odds in Congress
The Ban Artificial Superintelligence Act faces steep obstacles to passage in a Republican-controlled Congress, where lawmakers have generally favored a lighter regulatory touch on AI development in order to maintain U.S. competitiveness against China. No companion Republican support for the measure has been reported.
The debate underscores a broader divide in Washington over how aggressively to regulate a technology that both parties agree carries significant national security stakes, but disagree sharply on how to manage.


Add comment