AI is being described as a weapon-grade technology that private companies should not control, a warning that is bad news for the AI industry's current hands-off setup.

• The argument: AI is closer to a super weapon than a normal business, and governments, not startups, usually hold the keys to that kind of power.
• Trump is letting Dario Amodei at Anthropic and Sam Altman at OpenAI run this tech freely, which is unusual — nobody would hand Mark Zuckerberg nuclear weapons.
• The comparison to nukes rests on scale of damage: a mistake could let a hostile regime build a bioweapon.
• Private firms are also effectively deciding how many people lose their jobs, a call normally made by governments.
• The expectation is that this realization hits hard next year, which would mean pressure for government control over the leading labs.

Outlook: Expect louder calls for Washington to take direct control of frontier AI, which would be a direct threat to the independence of the biggest AI companies.