Bernie Sanders says AI danger is "probably greater" than nuclear weapons and proposes 20-year prison terms for violations

Sen. Bernie Sanders plans to introduce legislation next week that would permanently ban super intelligence - AI systems smarter than humans and capable of escaping human control - and pause advanced AI development. Sanders, speaking ahead of his appearance at the Pro-Human Assembly conference on AI, said the potential danger of artificial intelligence is "probably greater" than the threat posed by nuclear weapons. Violators of the proposed law would face 20-year prison sentences.

The proposed penalties and the nuclear comparison

Sanders told CBS News that stiff penalties are justified when dealing with technology that can threaten humanity's future. When asked about the 20-year sentence matching the penalty for illegal nuclear weapons development, Sanders said "the analogy is appropriate, exactly."

"You want to develop a nuclear weapon that could threaten humanity, you should be punished," Sanders said. "If you are moving forward on an AI development that could threaten humanity, you should be punished."

He went further, directly comparing the two threats. "No, I think AI, the potential danger of AI, is probably greater than nuclear weapons," Sanders said.

Calls for a slowdown gain bipartisan attention

The Pro-Human Assembly conference drew both Democrats and conservative Republicans, including Rep. Chip Roy and Sen. Marsha Blackburn. House Speaker Mike Johnson said he expects top AI company leaders to meet with President Trump at the White House "within the next week."

The political attention follows the public resignation of AI researcher Jacob Coxon from Anthropic, the company behind the Claude AI app. Coxon warned that while current AI is safe, the technology could one day threaten humanity as it grows more powerful. Several other Anthropic employees agreed with his assessment. Anthropic CEO Dario Amodei subsequently called for a slowdown at frontier AI labs, a position quickly echoed by OpenAI CEO Sam Altman and Elon Musk.

President Trump dismissed these concerns, calling them a "HOAX" and arguing that a strong president is all that's needed. Sanders criticized that stance directly. "I think when you have a president, Mr. Trump, who talks about this as a hoax, I think it is embarrassingly stupid for the leader of this country to hold that position," he said. "You listen to the people who know the most, and you act accordingly."

Government intervention and the limits of self-regulation

Sanders acknowledged he uses AI for research and recognizes its benefits, particularly in healthcare. But he argued that tech CEOs cannot be trusted to regulate their own industry.

"Look, let's be clear - they are pushing forward and have pushed forward for years to make as much money as they can," Sanders said. "And they have done very well. These are all the wealthiest people on the planet. And I think in a democratic society you're going to have to have government intervening very strongly."

The tension between rapid commercial development and safety concerns has become a defining debate in generative AI and LLM circles. Frontier labs continue pushing capabilities forward while a growing number of researchers inside those same labs warn about losing control of the systems they build.

Why this matters for IT and development professionals

Legislation targeting super intelligence and advanced AI development would directly affect the tools and systems IT teams build and deploy. A pause on frontier development could slow the release of new models that developers currently integrate into enterprise applications. The proposed criminal penalties - 20 years in prison - would create personal legal risk for engineers and technical leaders working on systems that regulators later deem too advanced or dangerous. For professionals in AI for IT and development roles, the bill signals that compliance and safety documentation may soon become mandatory parts of the development lifecycle, not optional checkboxes. The bipartisan nature of the Pro-Human Assembly also suggests that AI regulation is one of the few areas where legislation could actually move through a divided Congress.