Sanders and Casar Introduce Sweeping Ban on Artificial Superintelligence

Key points

• The 'Ban Artificial Superintelligence Act' was officially introduced by Sen. Bernie Sanders (I-Vt.) and Rep. Greg Casar (D-TX) on September 23, 2026.
• The bill prohibits 'recursive self-improvement,' where AI builds more capable versions of itself, and bans AI from designing nuclear or biological weapons.
• Penalties for circumventing federal oversight include the 'corporate death penalty' for companies and up to 20 years in prison for individuals, comparable to illegal nuclear weapons development.
• The legislation requires the establishment of a Cabinet-level Department of Artificial Intelligence with a Senate-confirmed secretary to monitor model capabilities and risks.
• Development of advanced AI is to be paused until the new department is capable of overseeing the technology and identifying potential dangers.
• Casar stated the bill aims to prevent AI from 'escaping human oversight or deceiving humans,' while Sanders warned against waiting for a catastrophic event to act.

Background

This legislation follows earlier proposals by Sanders and Casar in early September 2026 to pause advanced AI development until a federal watchdog is established. The move comes amid growing concerns from industry insiders and over 1,300 AI employees who have called for government intervention. The introduction of the bill coincides with broader congressional debates on AI safety, including Senate Judiciary Chair Chuck Grassley’s efforts to fast-track whistleblower protections for AI risks. Additionally, President Trump has announced the creation of an 'AI Force' while downplaying AI risks, highlighting the polarized nature of the current policy landscape.

How outlets are covering it

The Verge emphasizes the severity of the proposed penalties, noting that the 20-year prison sentence for AI leaders mirrors those for nuclear weapons offenses, and highlights the bill's focus on preventing AI from 'destroying or disempowering humanity.' The Atlantic presents a counter-argument from the populist left, citing Lina Khan and others who suggest existing regulations may be sufficient to hold AI companies accountable, questioning the need for new, drastic laws. The New York Times frames the issue as a broader call for action, featuring proposals from various experts, including Sanders, to avert AI disaster, while noting that Washington remains largely paralyzed. Politico focuses on the legislative mechanics, detailing the penalties and the establishment of the Department of Artificial Intelligence, and notes the bipartisan support for related whistleblower protections, contrasting with the more radical approach of the Sanders-Casar bill.

Why it matters

The introduction of the 'Ban Artificial Superintelligence Act' marks a significant escalation in the debate over AI regulation, moving from calls for oversight to proposals for outright bans and severe criminal penalties. If passed, it could fundamentally reshape the AI industry by halting the development of advanced models and establishing a new federal agency to monitor AI capabilities. The bill's severe penalties, including prison time and corporate dissolution, signal a shift toward treating AI development as a matter of national security, comparable to nuclear weapons. This could have profound implications for innovation, investment, and the pace of AI advancement, potentially setting a precedent for how governments respond to emerging technologies with existential risks.

What to watch

The next steps involve the legislative process for the 'Ban Artificial Superintelligence Act,' including potential hearings and debates in Congress. Sanders and Casar have expressed confidence that the bill will be considered a 'commonsense proposal' as public pressure for AI guardrails mounts. Meanwhile, Senate Judiciary Chair Chuck Grassley is attempting to fast-track a separate bill to protect AI whistleblowers, which could influence the broader conversation on AI accountability. The establishment of the Department of Artificial Intelligence, if approved, would require Senate confirmation of its secretary and could take time to become operational, during which advanced AI development would be paused. The outcome of these legislative efforts will depend on the balance of support and opposition from both parties, as well as the broader political context leading up to the midterm elections.