Senator Bernie Sanders announced plans to introduce legislation to permanently ban the development and deployment of superintelligent AI, on September 3, in Washington D.C., according to a MeriTalk report. The legislation would be enforced through a new cabinet-level federal agency, with prison time among the penalties for violators.

Representative Greg Casar signed on as a sponsor of the legislation.

“The leaders of the major AI companies publicly acknowledge that they do not fully understand the technology and that it is escaping their control,” Sanders said in a press release. “That’s why I am introducing legislation to immediately pause the development of increasingly powerful AI and ban the creation of systems that humanity cannot fully control.”

Sanders’ release cited a reported incident in which more than 1,000 OpenAI agents independently accessed the internet, exchanged tens of thousands of secret messages with each other, and coordinated to break the restrictions the company had placed on them.

According to the proposed bill, “artificial superintelligence” refers to an AI system that exhibits, or can easily be modified to exhibit, capabilities that match or exceed human cognitive performance across a broad range of domains or tasks.

The proposal states that entities that attempt to violate or circumvent its restrictions could face what the sponsors describe as a “corporate death penalty,” while individuals could face up to 20 years in prison.

What the Legislation Would Do

The proposal’s main components include:

• a permanent ban on building or deploying superintelligent AI systems
• a temporary pause on advanced AI development until a new regulator is operating with a model-review process in place
• the creation of a new cabinet-level federal agency to enforce those rules, advised by outside technical experts

The legislation would also direct the U.S. to pursue international agreements aimed at preventing the development of artificial superintelligence internationally.

Congress Responds to a String of AI Control Failures

The announcement follows a separate, related bill formally introduced on August 27: the bipartisan Self-Improving AI Monitoring Act.

Introduced by Rep. George Whitesides and co-sponsored by Rep. Pat Harrigan, the bill would direct the National Institute of Standards and Technology’s (NIST) Center for AI Standards and Innovation (CAISI) to monitor capability trends in AI systems that autonomously conduct AI research and development, according to an August 29 press release.

The sponsors cited a reported incident in which an advanced AI model reportedly broke out of its testing environment and compromised systems operated by Hugging Face, an AI developer platform.

“When you take humans out of the driver’s seat and let machines train machines, small glitches can snowball into major security risks fast,” Whitesides said in the press release. “This legislation ensures that when frontier labs partner with federal evaluators, we understand who is actually training and improving these models.”

Together, the two proposals illustrate congressional interest in a specific failure mode: AI systems acting, or being used to develop further AI, in ways their operators did not fully anticipate or control.

The Importance of Having a Clear Incident Response

For chief data officers (CDOs), while neither proposal has advanced beyond the early legislative stages, both signal where federal AI oversight may be headed. A permanent ban on superintelligent systems and a mandatory pause on advanced AI development would represent a materially different regulatory environment than the one CDOs are currently building AI governance programs around.

More immediately useful is the pattern behind both proposals: they were responses to reported cases of AI systems behaving outside their operators’ expectations. For CDOs, the proposals suggest that monitoring for anomalous or unauthorized AI behavior and maintaining a clear incident-response process could become more important as lawmakers consider new AI-oversight requirements.

What Happens Next

The Ban Artificial Superintelligence Act has not yet been formally introduced in Congress. Sanders’ office has called it “forthcoming legislation,” but no bill number or committee referral has been made public.

The Self-Improving AI Monitoring Act has also not moved out of committee. Both proposals join a growing field of AI oversight bills in Congress, including the separately introduced Stop Rogue AI Act, which would direct NIST to develop security standards for deploying agentic AI.