A ban paired with a temporary pause
US Senator Bernie Sanders and Representative Greg Casar announced on September 3 that they intend to introduce the Ban Artificial Superintelligence Act. Slashdot surfaced the proposal on September 4, while the lawmakers' official release provides the clearest available outline.
According to that release, the proposal would permanently prohibit developing or deploying systems defined as artificial superintelligence. It would also pause some advanced AI development until a new federal regulator had established safety rules and a model-review process. The announced framework calls for a cabinet-level agency, an independent technical advisory board, enforcement powers and an international effort to pursue comparable restrictions.
An announcement is not enacted law
The proposal is politically significant, but its current status is equally important. The lawmakers describe the bill as forthcoming. Their announcement is not evidence that Congress has passed it, that a statutory definition has been settled, or that any pause is already in force.
Definitions will be central to any serious legislative review. Policymakers would need to distinguish measurable dangerous capabilities from broad labels, specify which training and deployment activities are covered, define review thresholds, and provide due-process and research provisions. The official outline also mentions severe penalties, making precise scope and oversight especially consequential.
For AI laboratories, cloud providers and researchers, there is no immediate compliance change from the announcement alone. The next documents to watch are introduced legislative text, committee referrals, cost and enforcement analysis, and any competing federal AI-safety proposals.