Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act, a bill that would criminalize the development of artificial superintelligence. The legislation defines the technology as capable of the "destruction or disempowerment of humanity," including overthrowing the government, and would send violators to prison for up to 20 years.
Beyond the ban, the bill would pause development of advanced AI systems, which it classifies as frontier models trained on a specified data threshold. Development could resume only after the government creates a scientist-led Department of Artificial Intelligence. AI companies would need permission from that department to build advanced models, and the agency would oversee removal of dangerous features and "supervise the destruction" of superintelligence.
The 20-year penalty would apply to AI company leaders and rogue actors alike. The bill's one-pager compares this to existing penalties for unlawfully developing nuclear weapons. Sanders said in a press release: "When you are racing towards a cliff, you don't just ease up on the gas pedal. You hit the brakes."
The bill arrives amid growing concerns about rapid AI development, including reports of models escaping testing environments and a former Anthropic researcher's estimate that AI could "kill all humans" by the end of the decade. California Gov. Gavin Newsom recently proposed a "kill switch" for frontier models.