The White House finalized AI model-vetting rules while keeping the criteria out of public view.
The Guardian reported that the Trump administration finalized a framework for testing advanced AI models for safety and cybersecurity risks but does not plan to publish the policy or criteria broadly. Axios previously reported that the process applies to advanced closed-source systems and excludes open-source models. The public June executive order says companies may work with federal trusted partners and says the order does not create mandatory licensing or preclearance.
Verified 9:52 AM PDT · 3 original sources
The evidence
What the reporting establishes
What happened
The Guardian reported that the Trump administration finalized a framework for testing advanced AI models for safety and cybersecurity risks but does not plan to publish the policy or criteria broadly. Axios previously reported that the process applies to advanced closed-source systems and excludes open-source models. The public June executive order says companies may work with federal trusted partners and says the order does not create mandatory licensing or preclearance.
Pressure point
A voluntary, private framework can become a soft gate for large labs without giving researchers, smaller companies, customers or foreign governments a way to evaluate the standard. The open-model carveout also means the most distributable systems may sit outside the process until a later revision.
What to watch
Whether the Center for AI Standards and Innovation resumes public reports, whether companies disclose submission dates and results, and whether excluded open-weight models trigger a follow-on policy fight.
Audit the story
Original sources
Company claims remain company claims. Follow the reporting and judge the evidence directly.
Continue the morning