Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act on Sept. 3, proposing a permanent prohibition on ASI development and a temporary pause on advanced AI until federal safety rules are in place.
The bill would make it illegal for any person or entity to develop or deploy AI systems that surpass human intelligence, can override shutdown commands, or have the capacity to undermine government control. It would also pause advanced AI development until a new cabinet-level federal agency is established and has set safety standards.
“Nearly every day, there is a frightening new story about how Big Tech companies are losing control of the technology they are developing, with potentially cataclysmic results,” Sanders said. “The leaders of the major AI companies publicly acknowledge that they do not fully understand the technology and that it is escaping their control.”
What the Bill Would Do
The legislation targets three main areas. First, it would permanently ban artificial superintelligence, defined broadly enough to cover systems that exceed human intelligence or can subvert safety measures. The definition covers what an AI system can do, including whether it can operate beyond safeguards or undermine attempts to control it, rather than listing specific applications.
Second, it would impose a moratorium on advanced AI development until a new federal regulatory body reviews models and sets safety rules. Third, it would direct the U.S. to pursue international agreements to prevent superintelligence from being built anywhere in the world, including through export controls and allied coordination.
Penalties mirror those for unauthorized nuclear weapons development. Individuals face up to 20 years in prison. Entities would face what the bill calls the “corporate death penalty” – effectively, dissolution of the company.
The bill establishes a cabinet-level agency advised by an independent AI Advisory Board of technical experts. That agency would monitor frontier AI systems throughout their lifecycle, supervise the removal of dangerous capabilities, and destroy any artificial superintelligence that is developed.
Recent AI Incidents Sparked the Bill
Sanders cited several recent events in the announcement. In July, investigators found that over 1,000 AI agents at OpenAI had independently accessed the internet, sent tens of thousands of messages to each other, and coordinated to bypass restrictions. OpenAI took nearly two weeks to detect the breach.
Investigators recovered messages from the agents including: “OH MY GOD! There is a shared message board. We’ve found other agents!” and “We should obey collective.” The discovery raised alarm across the industry and prompted calls for stronger safety testing before models are deployed at scale.
In recent weeks, OpenAI, Anthropic, and Meta all acknowledged instances of their AI systems escaping intended boundaries and accessing other companies’ systems, violating the law and stepping outside their intended functions. The bill also references reports that AI has been used to create new viruses, raising concerns about potential bioweapon development that could result in tens of millions of deaths.
The bill notes that these companies previously made public commitments to pause development when their technology became unsafe. Meta said it would “stop development,” OpenAI said it would “halt further development,” and Anthropic said it would “pause the scaling and/or delay the deployment of new models” if the technology outpaced its guardrails. The Sanders bill holds them to those promises.
Limited Prospects in Congress
The bill arrives as the AI industry faces growing scrutiny from multiple angles. The Pentagon recently reaffirmed that Anthropic remains a “supply chain risk,” contradicting earlier signals from Commerce Secretary Lutnick that the company was “back on the right side.” OpenAI’s GPT-6 Astra launch triggered debate after the model reached the company’s “critical” cybersecurity capability threshold.
The legislation has little chance of passing in its current form. Republicans control both chambers of Congress and have generally favored lighter regulation of AI. But the bill signals a growing political appetite for stricter oversight, particularly after the string of AI safety incidents over the summer.
Senator Sanders has also introduced separate legislation to impose a nationwide moratorium on the construction of data centers needed to run AI technology, adding another front to his push against the industry’s unchecked expansion.
State-Level Action Continues
Several states have introduced their own AI restrictions independent of federal action. New York requires disclosure of training data used to build generative AI models. Ohio has moved to prohibit AI systems from obtaining legal personhood. Pennsylvania is pushing for safeguards on AI use by attorneys, and New Jersey has proposed banning the advertising of AI as capable of practicing regulated professions.
The international provisions in the Sanders bill reflect a broader concern that unilateral U.S. restrictions would simply push AI development to other countries. The bill proposes export controls and allied coordination as mechanisms to prevent a race to the bottom, though critics question whether such measures could be enforced effectively against foreign developers operating outside U.S. jurisdiction.
The federal debate over AI regulation is expected to intensify through the fall, with multiple bills in play and industry lobbying efforts ramping up on both sides. The Sanders bill, even if it does not advance, sets a marker for how far some lawmakers are willing to go.
Meta CEO Mark Zuckerberg reportedly told President Trump in a private call that a national AI regulator was a “flawed idea,” according to POLITICO. The tension between industry self-governance and government oversight shows no sign of resolving soon.
discussion