Senators propose banning super-smart AI, with up to 20 years in prison

Sen. Bernie Sanders and Rep. Greg Casar have introduced a bill that would permanently ban the development of artificial superintelligence in the US. People who recklessly break the rules could face up to 20 years in prison. The bill is still in committee and is not law. ChatGPT, Claude and Gemini are not affected today.

What the bill says

The Ban Artificial Superintelligence Act of 2026 was introduced in September. Sanders is an independent from Vermont. Casar is a Democrat from Texas. Artificial superintelligence means a theoretical AI that could outperform humans at most thinking tasks. The bill also covers systems able to plan and carry out actions that could destroy humanity or take away human control.

It would prohibit developing, deploying, acquiring or distributing such systems. It would also temporarily halt development of certain advanced AI models and restrict systems that show dangerous abilities.

A new federal department

The bill would create a cabinet-level Department of Artificial Intelligence. Companies would need government authorization to develop or distribute qualifying advanced models. Those models would need approval before release.

• Monitor advanced AI development and set safety requirements
• Inspect powerful models before they are released
• Order risky systems disconnected from the internet, changed or shut down

The penalties

Certain company decision-makers, and unaffiliated individuals, who recklessly violate the law could face fines and up to 20 years in prison. Companies could lose authorization to operate. Intellectual property and assets tied to violations could be seized.

What it means for everyday users

There is no reason to expect lost access to current chatbots. The bill has not passed Congress, and it places no new limits on consumer AI services. Even if it became law, it would not automatically ban every existing chatbot. Its focus is superintelligence, dangerous capabilities and advanced systems that meet certain technical tests.

The next generation is where it could bite. OpenAI, Google and Anthropic are working on AI that reasons better, finishes complicated tasks and acts with more independence. Qualifying future models would face federal review first. That could mean slower releases of more powerful tools, though the real effect would depend on how the law was carried out.

What happens next

Passage looks uncertain. The bill also calls for international agreements to stop superintelligence from being built outside the US. That could be a big hurdle, because the Trump administration has stressed AI development and American competitiveness in the AI race. For now, the bill puts one question in front of Congress: how powerful should AI be allowed to become?