Bernie Sanders wants to make building superintelligent AI a federal crime.
Sen. Bernie Sanders (I-VT) and Rep. Greg Casar (D-TX) introduced the Ban Artificial Superintelligence Act this week. The bill defines superintelligence as AI capable of the "destruction or disempowerment of humanity," including overthrowing the government, and it would ban anyone from developing it. Violators would face up to 20 years in prison. The bill would also pause development of more advanced AI systems generally, giving Congress room to catch up.
The timing is pointed. The Trump administration has been pushing federal agencies to swap "AI" for "superintelligence" in official documents, and this bill would make developing exactly that a federal crime. It reads as much like a jab at the administration's framing as it does an AI safety measure, and it marks a sharp break from how Washington has handled AI so far. Most federal proposals have focused on disclosure, licensing, or liability. A flat ban paired with 20-year prison sentences puts this bill closer to how the US treats bioweapons research than how it treats software.
Whether a ban like this is even enforceable is a fair question, given that AI labs themselves can't agree on what "superintelligent" actually means.