Sanders Introduces Bills to Ban AI Superintelligence, Regulate Agentic AI

What Happened

U.S. Senator Bernie Sanders has introduced two pieces of federal legislation aimed at banning the development of superintelligent artificial intelligence and establishing regulatory oversight of agentic AI systems. The bills were reported by Computer Weekly and represent one of the most direct legislative challenges to advanced AI development yet introduced in the U.S. Senate.

The Bills

The first bill targets superintelligent AI, defined broadly as AI systems that meet or exceed human-level cognitive performance across most or all domains. The legislation would prohibit the development of such systems within the United States. The second bill is directed at agentic AI, meaning AI systems capable of taking autonomous actions, executing multi-step tasks, and operating with limited human oversight. That bill would impose regulatory requirements on developers and deployers of such systems.

The two bills together represent a legislative effort to place hard limits on the upper boundary of AI capability while also creating a governance framework for AI systems already being deployed in commercial and enterprise settings.

Background

Sanders has previously spoken on economic disruption caused by automation and corporate concentration in the technology sector. The introduction of these bills follows a period of intensifying Congressional attention to AI policy. The U.S. has not yet passed comprehensive federal AI legislation, though the Biden and Trump administrations each issued executive orders addressing AI development and procurement. Several bills have been introduced across both chambers in recent sessions, with most focusing on transparency, liability, or sector-specific regulation rather than outright capability limits.

The term superintelligence, as used in the bill, refers to a threshold concept in AI research describing systems that would surpass human cognitive ability across all domains. Agentic AI, the subject of the second bill, is already in commercial deployment. Major technology companies including Microsoft, Google, and Salesforce have released or announced agentic AI products that can browse the web, write and execute code, manage files, and complete multi-step workflows with minimal human intervention.

Legislative Context

The bills arrive alongside intensifying debate in Washington over how to govern AI development. A separate report, cited in recent wire coverage, called on the U.S. government to consider extreme measures including military options to prevent China from achieving artificial general intelligence before the United States. That report reflected a contrasting policy instinct: that the U.S. should accelerate AI development for strategic reasons, not constrain it.

Sanders's legislation places him among a smaller group of policymakers who argue that the risks of unconstrained AI development outweigh the strategic or commercial benefits of moving faster. No co-sponsors were named in the initial wire report.

What It Means in Practice

If passed, the superintelligence ban would affect research programs at major AI laboratories including OpenAI, Google DeepMind, Anthropic, and Meta AI, all of which have stated goals that include developing highly capable general-purpose AI systems. Enforcement mechanisms and definitions of the capability threshold that would trigger the ban were not detailed in available wire reports.

The agentic AI bill, if enacted, would require regulatory compliance from a broad range of companies already shipping products. The specific agency designated to oversee agentic AI under the bill was not identified in available reporting.

What Happens Next

Both bills will be referred to the relevant Senate committee for review, where they will require committee approval before proceeding to a full Senate vote.