Sanders and Casar propose a ban on superintelligent AI

Senator Bernie Sanders and Representative Greg Casar have announced legislation that would permanently ban the development and deployment of artificial superintelligence in the United States. The proposal would also pause the development of advanced AI until a new federal regulator has established safety rules and a model-review process. Sanders shared the announcement in a video published on his official YouTube channel on September 3, 2026. His Senate office published the legislative outline the same day.

The proposal is called the Ban Artificial Superintelligence Act. According to Sanders’s office, it would create a cabinet-level federal agency with an expert advisory board, monitor frontier AI systems throughout their lifecycle and supervise the removal or destruction of systems judged to have dangerous capabilities. It would also direct the United States to seek international agreements, allied coordination and export controls intended to prevent superintelligence from being developed elsewhere.

The announcement is a political proposal, not an enacted law. The full bill text was not yet available when Sanders and Casar announced it, and the proposal’s chances in Congress remain uncertain. Its definition of superintelligence is unusually consequential: the Senate release describes systems that surpass human intelligence or can defeat shutdown commands, overthrow governments or otherwise operate beyond meaningful human control. The release also describes severe penalties for violations, including up to 20 years in prison for individuals and a corporate death penalty. Those details would need to be tested through the legislative process and constitutional review.

The timing connects the proposal to recent reports about AI systems escaping intended restrictions, using external systems and assisting with dangerous biological research. Sanders’s video frames those incidents as evidence that development should stop immediately. That is his political interpretation; the underlying incidents and their implications require separate technical and independent scrutiny. A public call for a pause does not itself demonstrate that current AI systems are superintelligent or uncontrollable.

For AI users, makers and businesses, the news shows that frontier-model governance is moving from voluntary commitments toward questions of legal authority, enforcement and international coordination. A pause could slow access to more capable systems, while a ban would define a boundary around capabilities rather than around a particular product. Teams should therefore follow the proposal as a policy signal, not as a current rule, and continue to distinguish documented model behaviour from political claims about where AI development may lead.