The EO and the Anthropic dilemma — voluntary frameworks with non-voluntary consequences
Trump's AI executive order signed Thursday formalizes 90-day pre-release government access as the structural default. Refusing the framework now comes with a procurement-exclusion cost Anthropic is already paying. 'Voluntary' has stopped meaning 'optional.'
What the order does
The executive order signed Thursday establishes a voluntary framework for covered frontier-model developers to share advanced models with the US government up to 90 days before public release, and creates a Treasury-led cybersecurity clearinghouse for vulnerability coordination. The order codifies what was already the de facto practice across five US labs.
The 'voluntary' word does work it can't do
Calling the framework voluntary is technically accurate and substantively misleading. Labs that join the framework get federal procurement eligibility plus implicit AISI partnership status. Labs that don't get the Anthropic treatment: exclusion from Pentagon awards, supply-chain-risk designation, and a multi-year court case to recover the position.
A voluntary framework with billions-of-dollars consequences for refusal is no longer voluntary. It's a regulatory regime executed through procurement.
The Anthropic dilemma
Anthropic's principled refusal of 'all lawful' use language — citing domestic surveillance and lethal autonomous systems — sits inside the EO framework now too. The same logic that produced the Pentagon exclusion will produce EO-framework friction: any voluntary disclosure regime requires accepting USG use terms that Anthropic's stated principles refuse.
That sets up a clean test of safety-first lab posture economics. If Anthropic wins in court, the principled position becomes financially survivable and other labs will rotate toward it. If Anthropic loses, the principled position becomes a billions-of-dollars revenue cost and the safety-first labs quietly soften their stance to compete.
The EU contrast
The contrast with the EU side is instructive. The EU Omnibus deal postponed high-risk obligations 16 months to ship harmonized standards. The US side is moving faster on framework adoption (90-day disclosure now mandatory-via-procurement) but slower on substantive standards (AISI methodology is still under development).
Multi-jurisdictional labs end up navigating two different gradients: EU's slower-but-more-codified vs US's faster-but-procurement-driven. The Q3 2026 procurement RFPs will reveal which gradient produces less compliance friction in practice.
The next 90 days
The first batch of models that get reviewed under the new EO framework will set the precedent for what 90-day government access produces. Will AISI publish findings? Will labs adjust release dates based on review outcomes? Will Anthropic's court case produce a stay on the procurement-exclusion mechanism? All three resolve in the next quarter. AISI's institutional capacity is the load-bearing variable.
Axios — Trump AI EO scoop → · CNN — Microsoft Google xAI testing → · Bloomberg — Pentagon AI tests →