The Trump administration is shifting from open AI growth to preview-first regulation. A proposed White House executive order would establish a working group to review new AI models before public release, a direct response to models like Claude Mythos and GPT 5.5. On Moonshots, Alex Gross noted these frontier models can discover system vulnerabilities faster than government agencies like the NSA, putting civilian capability ahead of national security for the first time.
"The tension lies between national security and the need for speed."
- Brian Elliott, Moonshots with Peter Diamandis
Silicon Valley is pushing back hard. On All-In, David Sacks dismissed talk of an "FDA for AI" as a federal power grab, arguing senior Trump officials do not support an approval regime. He contends the industry already self-polices through coordinated security tests and that cyber defense tools must reach the private sector quickly to counter threats. The IMF recently cited Claude Mythos's ability to exploit browser vulnerabilities as an example of AI-amplified cyber risk to the global financial system.
Regulatory momentum is building nonetheless. The European Central Bank’s Christine Lagarde is escalating rhetoric against private stablecoins, framing them as risks to monetary policy - a parallel fight over who controls new digital infrastructure. This regulatory pressure coincides with unprecedented consolidation. David Sacks cited Anthropic’s revenue tripling from $10B to $30B ARR between January and March 2024, then hitting $44B in April, putting it on a path to potentially become "the most powerful monopoly in history."
"He argues the real regulatory need is for cyber defense, as models like Mythos and OpenAI's equivalent give sophisticated hacking capabilities."
- David Sacks, All-In with Chamath, Jason, Sacks & Friedberg
The giants are bypassing traditional adoption channels. OpenAI’s $10B venture with TPG and Anthropic’s $1.5B deal with Blackstone represent a new distribution model, using private equity to mandate AI integration across thousands of portfolio companies. Salim Ismail calls this the "organizational singularity." The infrastructure race is also narrowing. Demis Hassabis admitted even Google is compute-constrained, choosing which frontier models to build based on hardware availability.
With AI capabilities now a national security concern, the clash is set: speed versus oversight, monopolization versus competition, and whether the government or Silicon Valley decides what gets built.
