Proposals would prohibit superintelligent AI, pause covered advanced development and give federal regulators greater authority over powerful systems.
United States | Artificial Intelligence and Public Policy
Information checked on 7 November 2026.
An AI superintelligence ban is gaining attention as some Democratic lawmakers push for stronger federal oversight of the most powerful artificial intelligence systems.
The proposals include a permanent prohibition on artificial superintelligence, a temporary pause on covered advanced AI development and the creation of a dedicated federal regulator.
At the centre of the debate is the Ban Artificial Superintelligence Act of 2026, led by Representative Greg Casar, a Texas Democrat, and Senator Bernie Sanders, an independent from Vermont.
A separate proposal from Democratic Senators Michael Bennet and Peter Welch would establish an independent Federal Digital Commission. Although both approaches seek stronger oversight, their structures and restrictions differ.
Neither proposal should be understood as a new ban already taking effect.
Who Is Calling for the AI Superintelligence Ban?
Casar and Sanders announced their legislative plans in September, before the November midterm elections.
The Senate version, S. 5493, was introduced on 23 September 2026. The House version, H.R. 10538, followed on 24 September. The official records reviewed list their introduction and referral to congressional committees.
Calls for action have continued since the election.
Representative Yassamin Ansari, an Arizona Democrat and an original House cosponsor, urged Democrats on 4 November to prioritise restrictions on advanced AI development, industry regulation and a AI superintelligence ban.
The renewed attention reflects growing interest in AI oversight among lawmakers preparing their priorities for the next Congress. However, support for stronger regulation does not mean every supporter favours the same permanent prohibition.
What Would Count as Artificial Superintelligence?
The bill defines artificial superintelligence through capabilities.
Its definition includes systems that exceed human cognitive performance across most domains or tasks. It also covers systems capable of planning and carrying out the destruction or disempowerment of humanity, including by undermining the federal government.
The proposed restrictions extend to specified superintelligence precursor characteristics.
Examples include the ability to resist shutdown, gain unauthorised access to protected infrastructure, substantially accelerate AI research or evade effective human oversight.
This makes the proposal broader than a restriction on systems already described by their developers as superintelligent. The capabilities a system possesses would matter more than its marketing label.
How the Permanent AI superintelligence ban and Temporary Pause Differ
The proposal contains two distinct mechanisms.
A Permanent Prohibition on Superintelligence
The legislation seeks to prohibit artificial superintelligence and covered precursor systems.
Its restrictions would extend beyond development and deployment to activities such as possession, funding, importation and transfer.
The intention is to prevent prohibited capabilities from being created, circulated or maintained.
A Temporary Pause on Covered Advanced AI
A separate provision would pause covered advanced AI development until the proposed department is fully staffed and has established the required safety rules.
The bill initially defines advanced AI using a training-compute threshold of at least 10²⁵ operations, subject to adjustment.
The pause would therefore operate through the legislation’s definitions and requirements. It is not written as a blanket instruction to switch off every AI application.
Nevertheless, the broader capability restrictions could affect systems beyond those commonly labelled superintelligent.
What Would the New Federal AI Regulator Do?
The Casar–Sanders bill would create a cabinet-level Department of Artificial Intelligence.
An AI advisory board would provide scientific and technical advice. The department would oversee covered systems throughout development and deployment, rather than relying only on checks immediately before public release.
Its responsibilities would include monitoring dangerous capabilities, supervising their removal and ensuring prohibited superintelligent systems are rendered inoperative.
Covered advanced systems would also require approval before deployment.
Supporters argue that this structure would place decisions about potentially catastrophic risks under public oversight. Their position is that voluntary commitments from developers are insufficient when the consequences could extend far beyond a company’s customers or shareholders.
What Penalties Does the Bill Propose?
The legislation includes substantial enforcement measures.
Certain reckless violations by covered decision-makers or unaffiliated actors could carry fines and prison sentences of up to 20 years.
Other provisions address industry employment restrictions, revocation of operating charters and forfeiture of assets connected with violations.
These are proposed penalties with specific conditions. They are not current punishments automatically applicable to ordinary AI development or use.
A Separate Proposal Would Create a Federal Digital Commission
Senators Michael Bennet and Peter Welch have proposed a different oversight structure through the AI Regulator Act.
Their proposal would establish an independent Federal Digital Commission with responsibilities covering frontier AI models and major digital platforms.
Its announced powers include reviewing models before release, requiring risk mitigation and incident reporting, and delaying certain releases that could pose catastrophic risks without adequate safeguards.
The commission could also impose civil penalties of up to 15% of a company’s previous-year global revenue.
| Issue | Casar–Sanders Proposal | Bennet–Welch Proposal |
|---|---|---|
| Proposed regulator | Department of Artificial Intelligence | Federal Digital Commission |
| Structure | Cabinet-level executive department | Independent commission |
| Main approach | Permanent superintelligence prohibition and a temporary advanced AI pause | Regulatory review, safeguards and intervention against serious risks |
| Broader coverage | Covered AI systems and their development | Frontier AI and major digital platforms |
The proposals share an interest in dedicated oversight, but they should not be presented as a single bill.
What the Proposals Could Mean for AI Businesses
If enacted, these measures could change how developers plan research, document safety decisions and release powerful models.
Approval requirements would make regulatory review an additional stage in product development. Ongoing monitoring could also require companies to maintain evidence about system behaviour after deployment.
For businesses purchasing AI services, the implications would depend on which suppliers and models fall within the final rules.
From a business perspective, several practical questions remain unresolved: how quickly reviews could be completed, how consistently capabilities would be assessed and how regulatory costs would affect smaller developers.
A well-resourced regulator could provide clearer expectations. An uncertain or slow process could complicate research and investment decisions. Those are possible consequences, not outcomes established by the introduction of a bill.
The Challenge of International Enforcement
The Casar–Sanders approach also calls for international agreements and coordination intended to prevent superintelligence development beyond the United States.
That ambition creates an enforcement challenge.
A domestic prohibition would not automatically bind overseas governments or developers. Its wider effectiveness would depend on international cooperation, verification and agreement about which capabilities should be restricted.
Policymakers would also need workable ways to distinguish permitted research from prohibited development as technology changes.
These questions make the design of enforcement as consequential as the headline commitment to a AI superintelligence ban.
What Happens Next?
The immediate issue is whether lawmakers can build support for a specific legislative approach.
A permanent AI superintelligence ban, a temporary development pause and a regulator with authority to review releases represent different policy choices. Negotiations could alter their scope, definitions and enforcement provisions.
For now, the proposals signal a push to give public institutions more control over decisions about powerful AI systems.
The next meaningful developments will be committee action, revised legislative text and evidence of broader support for whichever framework advances.
Explore more global business, technology and innovation coverage in The Empire Magazine’s article archive.
Read our related coverage: canva-layoffs-exit-package/
Follow The Empire Magazine:
Facebook | Instagram
The Empire Magazine | Crown For Global Insights
