US Lawmakers Propose Sweeping AI Restrictions Including Superintelligence Ban

Legislation Seeks to Pause Advanced AI Development

I've been following the rapid rise of artificial intelligence and its potential risks, especially as industry insiders issue increasingly dire warnings. Recently, a new piece of legislation introduced by Senator Bernie Sanders of Vermont and Representative Greg Casar of Texas is drawing attention for its sweeping restrictions on the AI sector — including an outright ban on developing what's known as artificial superintelligence.

The nearly 20-page bill, unveiled Wednesday, is designed to halt development of AI systems that can outperform humans across a broad range of domains. The legislation also proposes establishing a federal regulatory agency to oversee frontier models and ensure they don't become dangerous. This includes the power to remove and destroy any such models deemed harmful.

"It's unlikely to pass but it's courageous and much safer than the status quo," said Alex Turner, a former Google DeepMind researcher who resigned in June.

The Proposed Superintelligence Ban

The heart of the legislation centers on a ban against creating AI systems with cognitive capabilities that exceed or match human performance. The language is precise, stating that these systems must not have the power to "destroy or disempower humanity, including by overthrowing the federal government." This definition, while ambitious, reflects growing concerns within the research community about how quickly AI development is progressing.

Some experts argue the definition may be too broad and could inadvertently target current AI systems that are already quite advanced. However, others believe it's a necessary precaution given the potential for catastrophic outcomes.

Penalties and Enforcement Mechanisms

Violations of the proposed superintelligence ban would carry severe consequences — up to 20 years in prison and a 10-year ban from working in the AI industry. For companies, the bill introduces what it calls a "corporate death penalty," where charter holders could lose all intellectual property and assets to federal seizure.

The legislation also seeks international coordination to implement a global ban on superintelligence development. That would involve pursuing treaties and export controls on computing infrastructure used for AI development — a move that highlights the transnational nature of the threat posed by advanced AI systems.

Contextualizing the Bill

This effort comes amid mounting pressure from researchers who are publicly voicing fears about AI's trajectory. Earlier this month, Jacob Coxon, a former AI researcher at Anthropic, went viral after announcing his resignation with a social media post detailing how serious many in the field are about AI risks.

"The people building AI earnestly believe that it could kill us all by the end of the decade," he wrote. These sentiments echo those shared by other researchers who have since echoed concerns about the lack of oversight and the rapid pace of innovation in the field.

The Political Landscape

While the bill is ambitious, it's not without political hurdles. The legislation faces an uphill battle in Congress, particularly given that many AI companies are powerful stakeholders with deep pockets. Yet, the fact that it was introduced by two prominent figures — Sanders and Casar — suggests that there's growing momentum behind the idea of taking action before it's too late.

The political dynamics shift further when we consider that President Donald Trump recently told the UN General Assembly that he intends to rename AI as "super intelligence" in all US government documents. In his remarks, he dismissed any notion of a global agreement to control AI, calling such efforts a "globality scheme." His comments highlight the broader tension between regulation and innovation, even within the highest levels of government.

Why This Matters Now

With companies like OpenAI and Anthropic preparing for public offerings in 2027, the timing of this bill is significant. As these firms prepare to go public, there's a critical window for lawmakers to set boundaries that protect society from potential harm.

The bill is not just about stopping AI development — it's about ensuring responsible innovation. It proposes a framework where progress can continue, but with safeguards in place to prevent catastrophic misuse or unintended consequences. In this light, the legislation represents more than a policy proposal; it's an attempt to define the limits of human control over technology that might otherwise surpass us.

Conclusion: A Necessary Caution

While we may not see the bill pass anytime soon, its introduction signals a crucial conversation happening across American politics. We're at a crossroads where the speed of technological advancement outpaces regulatory frameworks. This legislation offers one possible path forward — a framework that demands careful oversight, accountability, and a recognition that artificial intelligence isn't just another tool but a force that requires thoughtful governance.

As we continue to track developments in AI policy, I'll be keeping a close eye on how this bill evolves and whether other lawmakers follow suit. The stakes are high, and the need for action is urgent — not just for the future of technology, but for the future of humanity itself.