Bernie Sanders and Greg Casar announced the Ban Artificial Superintelligence Act proposal on September 3, 2026. It would pause advanced AI development until a federal regulator establishes safety rules and a model-review process, permanently prohibit artificial superintelligence as the sponsors describe it, and create a federal agency. The proposal is not enacted law; individuals could face up to 20 years in prison under its proposed penalties.
What Sanders and Casar announced
The proposal would place a temporary pause on advanced AI development while a new regulator sets safety rules and reviews models. It would also permanently prohibit developing and deploying artificial superintelligence, which the sponsors describe in terms including systems that surpass human intelligence or can subvert shutdown commands.
Sanders frames the policy as a way to put decisions about AI’s future in the hands of the public and its representatives rather than a small group of technology executives. In his public argument, he calls for both an immediate pause on advanced AI and a permanent ban on superintelligence.
How the proposed rules would work
The measure would establish a cabinet-level federal AI agency advised by an Artificial Intelligence Advisory Board. The agency would monitor frontier AI systems, oversee the removal of dangerous capabilities and supervise the destruction of prohibited artificial-superintelligence systems.
The proposed penalties are severe: individuals could receive up to 20 years in prison for violating or circumventing the proposed prohibitions or pause. Entities could face what the sponsors describe as a “corporate death penalty.”
Why the proposal relies on international coordination
A U.S. measure would be domestic, so the proposal also calls for the United States to pursue international agreements, coordinate with allies and use export controls to prevent superintelligence development worldwide. Those steps are part of the proposed approach, not a guarantee that development elsewhere would stop.
Sanders has also argued for international cooperation on AI regulation. In a September 10, 2026 interview, he discussed the need for congressional action and coordination beyond the United States.
The debate over a pause and a permanent ban
AI researcher Gary Marcus supports a temporary pause and a federal AI agency, but opposes a permanent unilateral ban on research into superhuman AI. He argues that such a ban would be too broad and could leave the United States behind. His position draws a distinction between pausing development while regulation improves and permanently prohibiting the research itself.
