Key Takeaways

  • Senator Bernie Sanders is floating a legislative proposal that imposes up to 20 years in prison and a corporate death penalty for building unapproved artificial superintelligence.
  • The legal proposal defines superintelligence as any software matching human cognitive abilities across a broad range of tasks, which directly mirrors the stated mission of leading AI labs.
  • Jordi Hays notes that forced dissolution represents an extreme break from normal corporate wind-downs and bankruptcies.
  • John Coogan warns this policy creates an entrenched cartel of 10 approved companies where the eleventh builder gets shut down by the state.
  • Founders like Jamie Cox of Fluidstack and Will Manidis are pushing back with calls for clear, pro-growth guardrails instead of blanket prohibitions.

Defining Superintelligence as a Felony

Most founders expect tech policy to arrive as paperwork, reporting requirements, or compliance fees. The draft legislation associated with Senator Bernie Sanders takes a radically different route: criminal indictments and state-mandated corporate executions.

As John Coogan explains, the draft proposal sets extreme penalties: “Entities will shall shall be subject to the corporate death penalty and persons shall be shall be subject to not more than 20 years in prison if they don't pause AI development.” His summary captures the core prohibition: “His overall proposal is banning artificial super intelligence so no person or entity may develop or deploy super intelligent AI systems.”

The danger lies in how the text defines the crime. Coogan notes that the bill “defines artificial super intelligence as an artificial intelligence system that exhibits or can easily be modified to exhibit capabilities that match or exceed human cognitive performance and capabilities across a broad range of domains or tasks.” That is not a description of science-fiction doomsday machines. It is almost verbatim what frontier model companies print on their homepages as their core roadmap.

The 11th Company Problem and Forced Dissolution

Silicon Valley is used to seeing companies run out of cash and fail quietly. The legal mechanism in this bill is something else entirely. Hays points out the gravity of the language: “Corporate death penalty is a line that you don't hear a lot, right? Usually pe usually these companies just, you know, go bankrupt and wind down, but corporate death penalty.”

When a government claims the authority to dissolve a company for training software, the immediate outcome is regulatory capture. Incumbents who can lobby Washington will secure exemptions or compute licenses, while challengers face legal extinction.

Coogan outlines the trap: “The real you're going to see a lot of push back from people who are like the regulatory stuff is going to be like these 10 companies and I'm going to be number 11 and I'm basically getting the corporate death penalty then because I didn't make the cut.”

The Pushback from Growth-Focused Founders

Builders are organizing against this framing. Founders like Fluidstack's Jamie Cox and Will Manidis argue that top-down criminal bans confuse speculative future risks with practical technology policy. Instead of threatening engineering teams with federal prison, they advocate for predictable rules that keep domestic compute infrastructure growing.

When politicians define standard software targets as existential threats, they do not stop the code from running globally. They simply guarantee that early-stage teams without armies of lobbyists get locked out of the market entirely.

What to Do With This

Audit your product positioning and public roadmap before Q3. If your website claims you are building general intelligence across all domains, revise your copy to specify concrete business workflows and vertical agent capabilities that avoid the legislative crosshairs of broad frontier model definitions.