What AI oversight looks like if Democrats take Congress

As the midterms approach, Democrats appear to be on the rise, with recent race ratings and polling favoring them in several key races. The party needs to gain control of four Senate seats and three House seats to take over both chambers. President Donald Trump's approval rating has dropped to 32%, a 10-point decrease over the past year. In response to these political shifts, new legislation is being proposed that could significantly alter how artificial intelligence is regulated in the United States.

The Ban Artificial Superintelligence Act, introduced by Senator Bernie Sanders and Representative Greg Casar, calls for a complete ban on artificial superintelligence, which is defined as AI that surpasses human cognitive capabilities across most domains or poses a threat to humanity's existence. This bill would also impose an immediate pause on advanced AI development, followed by the establishment of a federal regulator to enforce safety rules and model-review processes.

Another proposed bill, the FRONTIER Act, introduced by Jay Obernolte and Lori Trahan, offers a more moderate approach. It would require model cards detailing AI capabilities and risks, implement risk-management frameworks, mandate incident reporting, and require ongoing assessments of advanced AI systems. The FRONTIER Act also emphasizes independent audits, where third parties evaluate the safety and alignment of new models.

Unlike the Ban Artificial Superintelligence Act, the FRONTIER Act includes a less stringent punishment system for corporate violations and proposes the creation of an independent oversight body.

Meanwhile, the Trump administration has taken a different approach, voluntarily establishing a framework that allows companies to provide federal agencies access to their models ahead of release. The administration has also renamed artificial intelligence as "superintelligence," rejecting international efforts to regulate its development.