Rep. Greg Casar (D-TX) announced he is introducing legislation with Sen. Bernie Sanders (I-VT) to ban the development of "superintelligent" artificial intelligence. The proposal seeks to prevent the creation of AI systems that the lawmakers say could surpass human control or possess the capability to develop weapons of mass destruction.
The announcement comes as debate continues over the pace of AI advancement and the role of government oversight. Casar stated that while some AI company executives have discussed a voluntary pause in development, a federal agency is necessary to enforce safety standards. House Speaker Mike Johnson (R-LA) has suggested that technology companies should lead AI safety efforts, a position Casar contested during a recent interview.
The bill text is scheduled for release in the coming days. Casar cited a poll by Data for Progress, stating that approximately 70% of Americans support an immediate ban on artificial superintelligence, while 25% are opposed. The congressman argued that the U.S. should lead in AI diplomacy and pursue international agreements, similar to Cold War-era nuclear nonproliferation treaties.
This legislation would affect technology companies and federal regulators by establishing a federal authority to oversee high-level AI development. While the specific number of companies impacted was not reported, the bill targets the creators of "superintelligent" systems. If passed, these developers would face prohibitions on certain types of advanced AI models rather than the current system of voluntary safety commitments.
For the general public, the impact would be felt through shifts in national security policy. Casar argued that without these regulations, there is a risk of losing human control over technology, which some experts have warned could lead to mass extinction. Conversely, Donald Trump has warned that a slowdown in U.S. development could result in the country losing its competitive edge to China.
The proposed ban could also impact international trade and diplomatic relations. Casar noted that because Chinese AI models often train on American versions, U.S. rules and export controls could slow development globally. The next step for the legislation is the formal release of the bill text. A specific date for a vote or the implementation of a new regulatory agency has not been established.