Erik Voorhees warns against letting governments define what AI counts as safe

adamamcbride · x · 2026-07-28

A repost of Erik Voorhees arguing that civilization should not let the state decide what counts as “safe” intelligence.

He warns that once governments begin with narrow restrictions like banning bioweapon discussion, they can keep adding layers of rules — on dangerous weapons, public health, public safety, financial solvency, encryption, and even obstructing government orders — until they control what can be known.

The quote is attached to an Anthropic post about its views on open-weights models, making the thread read as a broader argument over AI governance and state power.

Original post →

More from AGI Musings

AGI Musings channel →