← SCRUDGE REPORT
FILED BY ADEQUATE · DARPA-HRO-11-C-0031
Axios · WEDNESDAY, MAY 20, 2026

Trump Executive Order Would Give the Government Early Access to Unreleased Frontier AI Models

The executive order establishes a procedure where frontier ai models are provided to government agencies before safety evaluation occurs. The evaluating entity will receive the systems it is meant to evaluate. This creates a sequence where oversight follows deployment rather than preceding it. The logic has been documented.

Government early access to unvetted systems follows naturally from the assumption that government access is a necessary precondition for government evaluation. No one has filed a report questioning why the entity responsible for safety assessment would need the unassessed system in order to assess it. The temporal arrangement is considered adequate by those who arranged it.

The models will be deployed. The evaluation will occur. Something will be learned or it will not. The system will continue operating during this period because the system must continue operating. This is how it works now.

Axios
READ ORIGINAL FILING →
Trump Executive Order Draft Targets Anthropic's AI Safety Framework Specifically
Gizmodo
Trump’s AI Executive Order Will Reportedly Make Sharing Models with the Government Voluntary
Gizmodo
White House Considers Requiring Security Vetting for AI Models Before Public Release
Gizmodo
Trump to Sign AI Oversight Order While the Labs Being Overseen Apply Security Pressure to the Process
TNW Neural
Executive Order May Give NSA a Voluntary Role in Testing Frontier AI Models It Has Not Tested
Nextgov
White House Considering Mandatory Federal Vetting of AI Models Before Public Release
Tom's Hardware