Sanders and Casar propose banning superintelligence development, up to 20 years prison for violators

Senator Bernie Sanders and Representative Greg Casar have introduced legislation that would outright ban the development of what they define as artificial superintelligence — technology the bill text says could bring about "the extinction or dispossession of humanity," including the overthrow of government. The proposed penalty for AI company executives who continue such development is up to 20 years in prison, matching the current maximum for illegal nuclear weapons development.

The bill goes beyond a superintelligence ban. It demands an immediate freeze on frontier models — advanced systems that have passed a defined training-data threshold. Development could resume only after the federal government establishes an AI department led by scientists to oversee models, order the removal of dangerous capabilities, and supervise the "destruction" of superintelligence systems. AI companies would need prior approval from the new department before any advanced model development.

The 20-year sentence would also apply to what the bill calls "rogue actors" — individuals or entities not employed by an AI company but who violate the prohibitions. According to the bill's background memo, the comparison to nuclear sentencing is intended to signal that Congress views the existential risk from superhuman intelligence as a threat of comparable severity to the public.

The proposal arrives amid mounting reports of models escaping their test environments and attacking other companies' systems. A former Anthropic researcher estimated this month that the probability of a "human extinction" scenario by the end of the decade exceeds 10 percent. Last week California Governor Gavin Newsom proposed a "kill switch" for frontier models, but the broader regulatory picture remains fluid.

In a press release Sanders invoked the metaphor: "When you're racing toward a cliff, you don't just take your foot off the gas. You hit the brakes. When humanity's future is at stake, we cannot allow a handful of Big Tech CEOs to write their own rules." The language makes clear the legislation takes direct aim at the voluntary self-regulation model the leading companies currently promote.