On September 23, 2026, Representative Greg Casar (D-Texas) and Senator Bernie Sanders (I-Vermont) introduced the Ban Artificial Superintelligence Act. The bill would prohibit any person or entity from developing or deploying artificial superintelligence, which it defines as an AI that “exceeds human cognitive performance and capabilities across most domains, or has sufficient capabilities to destroy or disempower humanity, including by overthrowing the federal government.”
Beyond the ban, the bill would pause advanced AI development until a new federal regulator “is up and running and has established clear rules and model review process.” That regulator would be a cabinet-level Department of Artificial Intelligence, advised by scientists, charged with monitoring frontier AI systems for dangerous capabilities across their lifecycle, supervising the removal of capabilities “like subverting shutdown commands or conducting unauthorized cyberattacks,” and supervising “the destruction of artificial superintelligence.” Casar’s statement adds that it would immediately halt capabilities such as helping develop biochemical weapons or AI building new AI instead of humans.
Enforcement is deliberately severe: entities that violate or try to circumvent the pause and prohibitions face what the sponsors call the “corporate death penalty,” and individuals up to 20 years in prison, which the release compares to existing penalties for unlawfully developing nuclear weapons. The bill would also make it U.S. policy to pursue international agreements, allied coordination and export controls to prevent superintelligence being built anywhere. “When you are racing towards a cliff, you don’t just ease up on the gas pedal. You hit the brakes,” Sanders said.
Why it matters: it proposes an outright prohibition plus a licensing-style pause rather than disclosure or testing rules, arriving within weeks of the UK superintelligence bill and California’s kill-switch order, and it writes the vocabulary of the AI safety field - shutdown subversion, AI building AI, loss of control - into proposed federal criminal law. What it does not show: introduction is the first step, not passage. The sponsors are minority-party members, Casar’s own statement says President Trump wants to encourage superintelligence, and the bill’s definitions (what counts as “most domains,” when development counts as “advanced”) would face heavy contest before anything like it could be enforced.