Bernie Sanders and Greg Casar announced a plan for a temporary pause on advanced AI and a permanent ban on superintelligence. The full legal details are not yet clear.
New to this? Read it in simple words
- US Senator Bernie Sanders and Representative Greg Casar announced a plan about advanced AI. It includes a temporary pause on some advanced AI work.
- They say they want a permanent ban on AI that becomes smarter than humans. A new federal AI agency would review powerful models.
- The word superintelligence is strong, but it is not a clear legal test. A ban needs clear technical tests for AI that is too powerful.
- The full bill text was not public, and support in Congress was unclear. The proposal is not law and may never pass.
- Superintelligence
- An AI that would be far smarter than humans in almost every area.
- Bill
- A written proposal for a new law. Lawmakers debate and vote on it.
- Federal
- Of a country’s national government, not a state or local government.
What the lawmakers propose
The lawmakers say they want a permanent ban on AI that becomes smarter than humans, moves beyond human control, or can ignore commands to stop. They also want a temporary pause on some advanced AI work while new safety rules are written.
Their plan would create a new federal AI agency. Experts would advise it, and it would review powerful models during their full life. The lawmakers also want the United States to work with other countries and use export controls.
The sponsors want a temporary pause, a US regulator, a permanent ban, and work with other countries. This is not current law.
A ban needs a clear test
“Superintelligence” is a strong word, but it is not a clear legal test. An AI may be better than people at one task and much worse at another. A regulator must say which skills, levels of control, or dangerous actions would cross the line.
The same problem applies to a pause on “advanced AI.” The law must explain which models and training runs are included. Rules that are too broad may stop safe work. Rules that are too narrow may miss the systems they are meant to control.
The proposal starts a larger debate
At the time of the announcement, the full bill text was not public and support in Congress was unclear. The proposal is not law. It may change before lawmakers vote on it, or it may never pass.
Still, it asks an important question: who should decide when an AI system is too dangerous to build or release? Clear tests, public records, and independent checks will matter more than dramatic words.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · September 3, 2026Sanders, Casar to introduce legislation to ban artificial superintelligence and temporarily pause advanced AI development Office of Senator Bernie Sanders
- 2
We used the lawmakers’ official summary and independent reporting. We clearly separate the announced plan from current law. We do not guess how Congress will vote, and we explain that important technical definitions were still missing.