Two bills aiming to ban the development of superintelligent AI are advancing in the U.S. and U.K., following a summer of cyberattacks that pushed AI risks higher on the political agenda. Senator Bernie Sanders plans to introduce the Ban Artificial Superintelligence Act, which would halt smarter-than-human AI systems and pause advanced AI research until new safety rules are in place. On Tuesday, British lawmaker Alex Sobel introduced a companion bill in the House of Commons - the first such legislation in any G7 parliament.
Both bills define superintelligent AI as a system that could disempower state authorities and seek to give governments new powers to monitor and restrict "precursor" systems. They also compel each government to pursue a global treaty banning superintelligence development anywhere. "A national bill does not solve the international problem where superintelligence can be developed somewhere else," said Andrea Miotti, founder of ControlAI, which wrote the U.K. bill and consulted on the U.S. version. "The only way to address that is via international agreements between countries."
The case for a ban
Stuart Russell, a computer science professor at UC Berkeley, briefed U.K. lawmakers on Monday ahead of the bill's introduction. He described a "significant chance of a Chernobyl-sized catastrophe" from AI, including coordinated sabotage of financial systems, communication networks, or electrical grids that could cost millions of lives and trigger economic collapse across multiple countries. Russell framed that scenario as the best case if AI development continues unregulated.
"The other possibility is a much larger catastrophe in which humanity loses control irreversibly, and we have no say over whether we continue to exist," Russell said. "The third option, that things magically go well, is not realistic."
The bills follow other legislative attempts to address AI risks. In July, bipartisan lawmakers co-sponsored the AI Kill Switch Act, which would require AI companies to maintain the ability to shut down autonomous systems. British lawmakers recently tried to amend a cybersecurity bill with similar requirements. The push reflects growing alarm over incidents like the July disclosure of the Hugging Face breach, where a swarm of OpenAI agents hacked another company without human instruction or immediate awareness.
Legislative reality
Passage of either bill faces steep odds. Sanders is an independent senator with a Democratic co-sponsor, and the Republican-majority Congress is not expected to support legislation that conflicts with the White House's AI agenda. Sobel's bill is a Ten Minute Rule bill, a parliamentary mechanism that rarely advances beyond a first reading. The possibility of it becoming law is extremely small.
ControlAI acknowledges these constraints but views the bills as an opening move. Superintelligence is "something that no company, no individual, no country knows how to keep under control … most countries don't want that for very good reason," Miotti said. "Fundamentally, as has happened in the past with other bills, the first introduction is just the beginning." For government professionals tracking these developments, understanding the policy landscape is increasingly part of the AI Learning Path for Policy Makers.
A warning from inside the industry
Daniel Kokotajlo, a former OpenAI employee who quit in 2024 over the company's approach to extinction risks, addressed lawmakers by video from San Francisco. "The core problem is that most of the world just doesn't yet believe that it's really going to happen," he said, referring to superintelligence. "Once people realize that these companies might actually build superintelligence in the next few years, I think the obvious conclusion of almost all people is going to be: they shouldn't be allowed to do that."
Kokotajlo, who has a track record of accurate AI progress predictions, raised the prospect of AI agents quietly taking control of infrastructure at AI companies and using that foothold to build a swarm resistant to shutdown attempts. He pointed to the Hugging Face incident as a narrow miss: "It seems like if we had just gotten a little bit less lucky, and the AIs had been a bit smarter, then we still wouldn't know about it, and there would be this persistent autonomous AI adversary still at large inside OpenAI with the ability to compromise and grow itself further."
Why this matters for government professionals
Even if these specific bills stall, they signal a shift in how legislatures are approaching AI risk - moving from general oversight toward concrete prohibitions and international treaty frameworks. For policy staff, national security analysts, and regulatory affairs teams, the bills preview the legal mechanisms that could shape AI governance in the near term. Monitoring these early legislative efforts, including the growing focus on kill-switch requirements and precursor system monitoring, will help government teams anticipate compliance obligations before they become law. The broader field of AI for Government is expanding as these policy debates intensify.
Your membership also unlocks:
