The White House is trying to regulate AI without regulating AI, and the contradiction is louder than the policy.
The Summary
- White House Chief of Staff Susie Wiles says the administration won't pick winners in AI, while simultaneously the administration considers creating a working group to oversee new AI model releases
- White House officials have already briefed executives from Anthropic, Google, and OpenAI on oversight plans, signaling this is further along than just discussion
- Tech policy experts warn that executive-order-based oversight could slow innovation and should come from legislation instead
- This marks a reversal from Trump's deregulation stance, driven by pressure from national security officials worried about systems like Anthropic's Mythos
The Signal
The Trump administration is running two conflicting AI narratives at once. Chief of Staff Susie Wiles told reporters the government would avoid picking winners and losers, the kind of free-market language you'd expect from an administration that spent its early days gutting Biden-era AI safety rules. But behind closed doors, the White House is already briefing top AI companies on plans for a working group that would vet new models before release.
That's not avoiding picking winners. That's becoming the gatekeeper.
"Executive-order-based oversight could slow innovation and should come from legislation instead."
The timing matters. National security officials and tech analysts have been pressing the administration to consider risks from powerful AI systems, particularly after capabilities like Anthropic's Mythos raised new questions about what happens when model capabilities outpace safety protocols. The China competitiveness argument, which dominated early 2025, is losing ground to the "we need guardrails before something breaks" argument.
But here's the problem with doing this through executive order rather than legislation: it's policy by whim. One administration vets models, the next one doesn't. Tech policy experts quoted by Business Insider are already flagging this as a recipe for innovation slowdown without the legitimacy that comes from congressional action.
The bigger tell is who got the briefing. Anthropic, Google, OpenAI. The three companies already at the frontier. If you're not picking winners, why is the working group composition being negotiated with the incumbents? The companies with the most to lose from fast-moving competitors are the ones helping design the oversight structure. That's not neutral. That's regulatory capture before the regulation even exists.
Key contradictions:
- Free market rhetoric vs. pre-release vetting reality
- Deregulation stance vs. new oversight working groups
- Incumbent companies helping design the rules they'll be judged by
What's actually happening here is the White House trying to split the difference between the "move fast" wing and the "break things is terrifying" wing. They want to claim they're fostering innovation while also claiming they're taking risks seriously. The result is a policy position that satisfies neither and commits to nothing concrete.
The Implication
Watch who ends up on that working group. If it's dominated by Anthropic, Google, and OpenAI execs, you'll know this is about moat-building, not safety. If legislation doesn't follow within six months, you'll know this executive order approach is designed to be flexible enough to ignore when convenient.
For anyone building in AI right now, the message is clear: the rules are being written in real time by the people who already won the last round. If you're not in those rooms, you're not designing your future, you're inheriting someone else's.