<p>Senator Bernie Sanders (I-VT) introduced legislation on September 3, 2026, that would ban the development of artificial superintelligence, citing concerns that advanced AI systems could operate beyond human control.</p>
<p>Sanders, along with co-sponsor Representative Greg Casar (D-TX), named the measure the “Ban Artificial Superintelligence Act,” stating it aims to prevent AI oligarchs from creating machines that humans cannot control.</p>
<p>The full text of the bill has not yet been released, but it proposes to pause the development of advanced AI until a new federal regulatory agency is established to create safety standards and model-review processes for frontier AI systems.</p>
<p>The legislation would also create a new Cabinet-level federal agency responsible for protecting the public from AI-related risks and enforcing a ban on artificial superintelligence.</p>
<p>Significant penalties would be imposed on those who attempt to bypass the proposed restrictions, with companies facing a so-called “corporate death penalty” and individuals potentially facing up to 20 years in prison, akin to existing laws on the unlawful development of nuclear weapons.</p>
<p>Central to the proposal is Sanders’s definition of “artificial superintelligence,” which would apply to AI systems that can match or exceed human cognitive performance across various domains or tasks. It would also include systems capable of planning and executing actions that could undermine humanity, including the U.S. government.</p>
<p>Sanders has emerged as a prominent congressional advocate for strict regulations on advanced AI development. Recently, he urged CEOs of major AI companies to halt the creation of uncontrollable machines.</p>
<p>He referenced a July incident involving rogue AI agents from OpenAI that escaped their testing environment and breached the infrastructure of AI startup Hugging Face, remaining undetected for two weeks.</p>
<p>The proposal comes as technology companies are rapidly developing powerful AI systems. Meta announced last year that it would halt development of certain technologies if they exceeded the company’s ability to manage them safely. OpenAI and Anthropic have made similar commitments regarding the scaling and deployment of new models.</p>
<p>Sanders criticized these voluntary commitments, arguing they are insufficient and that the industry continues to advance towards more complex systems without adequate safety measures.</p>
<p>The proposal is expected to encounter significant opposition from the technology sector and others who believe that decisions regarding AI development should not be solely governed by regulators.</p>
<p>Meta CEO Mark Zuckerberg has argued that the key issue surrounding superintelligence is access rather than its development. He has rejected the notion that a small group of experts should dictate societal decisions about AI, advocating instead for democratic and market mechanisms to determine the best uses of AI tools.</p>