The man who brought you Space Force wants an AI Force—while the people building frontier models are begging someone to slow them down.
The Summary
- Trump announced plans for an AI Force and AI czar, promising the government won't "hinder or stifle" AI growth—while 20 nations called for human oversight at the UN (the U.S. wasn't one of them)
- An Anthropic researcher put odds of AI killing all humans in the next decade above 10%; CEO Dario Amodei, Sam Altman, and Elon Musk all asked government to slow frontier AI development
- Trump's December 2025 executive order directed DOJ to sue states like California and New York over AI safety laws, arguing state regulation gives China an advantage
- A counterterrorism scholar who's advised AI companies warns this approach will trigger an AI arms race and weaponization
The Signal
The timing tells you everything. While Anthropic's CEO is publishing essays asking for guardrails and his own researcher is putting double-digit odds on human extinction, the White House is announcing a military branch dedicated to not slowing down. That's not policy tension. That's policy incoherence.
Trump's AI Force proposal mirrors Space Force in structure—part of the Air Force, presumably focused on military AI applications. The details are thin, but the philosophy is clear: acceleration at all costs, wrapped in national security logic. The argument goes like this: China is moving fast, so any speed bump we add is strategic suicide. State-level AI safety laws become not prudent regulation but geopolitical self-harm.
That framing ignores what's actually happening in the industry. When Sam Altman and Elon Musk agree on something AI-related, you should pay attention. When they're both asking government to help slow frontier development, you're watching a rare moment of founder honesty. They know what's in the labs. They know the capabilities coming online in the next 18 months. And they're nervous.
"An Anthropic researcher put the odds that AI kills all humans within the next decade at more than 10%."
The state vs. federal fight is where this gets practical. California and New York passed laws requiring:
- Safety plans from the largest developers
- Incident reporting for serious failures
- Public disclosure of risks
Basic stuff. The kind of transparency you'd expect for any technology with civilization-level implications. Trump's DOJ is suing to block them. The legal theory is federal preemption plus the China excuse. The practical effect is a regulatory void right as capabilities cross into genuinely dangerous territory.
Here's what a counterterrorism scholar sees that the "move fast" crowd doesn't: militarizing AI development doesn't make you safer. It makes conflict inevitable. You create an AI Force, China creates one, then Russia, then everyone else with a defense budget. You've just guaranteed that the most powerful technology in human history gets optimized for winning wars, not for not killing everyone.
The Implication
Watch the state-level fights. If California's safety laws get struck down, you're left with voluntary commitments from companies racing each other to AGI. That's not a plan. That's hope.
The AI Force framing also changes the talent pipeline. Militarizing AI research pulls top researchers toward weapons applications and away from alignment, safety, and interpretability work. The smartest people end up building better targeting systems instead of figuring out how to make sure AI systems do what we actually want.
If you're building in this space, the regulatory certainty you wanted isn't coming from Washington. Look to the EU's AI Act for the global baseline. And if you're inside one of these labs, the gap between what leadership is saying publicly and what they're asking for privately should tell you something about the stakes.