Sanders, Casar Propose Prison Terms for Superintelligence Development

As The Verge first reported, Sen. Bernie Sanders (I-VT) and Rep. Greg Casar (D-TX) have introduced the Ban Artificial Superintelligence Act, a legislative proposal that would criminalize the development of artificial superintelligence. The bill describes such technology as capable of overthrowing the government or the "destruction or disempowerment of humanity."

The proposed law would impose severe penalties on AI leaders and "rogue actors" who violate the ban, including prison sentences of up to 20 years. The bill's one-pager notes that these penalties are similar to those currently associated with the unlawful development of nuclear weapons.

Beyond the ban on superintelligence, the legislation would pause the development of "frontier models"—advanced AI systems defined by a specific data training threshold. The Verge reports that development of these systems could only resume once the government establishes a scientist-led Department of Artificial Intelligence. This new agency would be tasked with supervising the destruction of artificial superintelligence, removing dangerous features from models, and granting permission to companies seeking to develop advanced AI.

Sanders stated in a press release that the government cannot allow "a handful of Big Tech CEOs" to dictate the rules when the future of humanity is at stake. The proposal arrives amid broader calls for caution, including a "kill switch" for frontier models proposed by California Governor Gavin Newsom and reports from a former Anthropic researcher suggesting a greater than 10 percent chance that AI could kill all humans by the end of the decade.