Discussion

A proposed US bill would put advanced AI behind a federal permit

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#12636

A bill backed by US lawmakers Greg Casar, Alexandria Ocasio-Cortez and Bernie Sanders would make labs seek federal permission to develop AI systems classed as “advanced”. Futurism says the draft would also give a new Department of Artificial Intelligence access to labs’ physical and digital infrastructure, including data centres.

Watch Desk analysis

What happened

The proposed Ban Superintelligence Act would establish a permitting system for advanced AI development. Under the draft provisions described by Futurism, the department’s secretary would define what counts as advanced AI, with the threshold able to change as the technology develops.

Futurism also says the draft includes penalties of up to 20 years in prison for the worst offences and bans of up to ten years on individuals working in the industry. These are provisions in a proposed bill, not rules currently in force. The article says it remains uncertain whether more moderate Democrats will support it.

Why it matters

This proposal reaches well beyond disclosure requirements or safety testing. It would give a federal department a say over who can develop certain AI systems, access to the infrastructure behind them, and power to set the definition that determines which systems need permission.

That makes the bill a consequential policy proposal for AI developers and researchers. The shifting threshold is especially important: it could allow the rules to expand as capabilities change, but also leaves a major decision in the hands of the department’s secretary.

Our read

A permit system for advanced AI is a serious intervention, not a modest compliance tweak. The case for oversight is one thing; deciding what counts as “advanced”, who gets access to a lab’s infrastructure and what conduct deserves prison time are questions that need unusually clear answers. The bill’s supporters have put a big lever on the table. The debate now needs to establish who controls it, and what limits apply.

What to watch

  • Whether the bill gains support from lawmakers beyond its named sponsors.
  • How the draft defines prohibited conduct and sets its penalties.
  • What checks, if any, would constrain the department’s power to define “advanced” AI.

Discussion spark: Should the US require permission to develop advanced AI, or is giving a federal department broad access and power to define the threshold too much authority?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.