Frontier-model researchers become subject to security-clearance requirements

Last updated

Tracker summary: The government contract requires security clearances for everyone working on the frontier models within two months.

No qualifying enforced clearance requirement covering a general frontier-model program has been verified.

At a glance

  • Assessment: Not Yet Testable
  • Confidence in assessment: 80%
  • Outcome: unresolved
  • Timing: pending
  • Evidence: insufficient
  • Predicted timing: May-July 2027
  • Primary source: AI 2027 archived scenario: May 2027: National Security, p. 14

What AI 2027 Predicted

Tracker summary: The government contract requires security clearances for everyone working on the frontier models within two months. The source is May 2027: National Security in the AI 2027 scenario.

Scenario source: archived version and page-number reference: May 2027: National Security, p. 14.

How We Track This

Require a binding US government contract, law or policy covering a named frontier model program, explicitly requiring formal security clearances for personnel working on its models. Record enactment date, personnel coverage, exemptions and the two-month compliance deadline. Implementation requires documented enforcement or completion figures, rather than an announced intention.

These operational thresholds specify this tracker’s test. Where the scenario uses qualitative language, they are interpretations rather than numbers quoted from the original.

Current Evidence

February 28, 2026: cleared deployment staff: OpenAI’s agreement announcement describes cloud-only deployment, cleared forward-deployed engineers and cleared safety/alignment researchers in the loop. This is cleared staff for specific deployments, not a clearance requirement for everyone working on the lab’s models. OpenAI agreement announcement

May 1, 2026: classified networks: The department announced agreements to deploy frontier AI capabilities in IL6 and IL7 environments. This is classified-deployment context, not evidence of a lab-wide personnel-clearance mandate or completion of the scenario’s two-month requirement. Department announcement

June 12, 2026 directive, lifted June 30, 2026: Anthropic reported a foreign-national access restriction on Fable 5 and Mythos 5 on June 12; its June 30 post says those export controls were lifted, with general Fable access returning July 1. This was an access restriction, not a formal security-clearance rule covering everyone working on frontier models. June 12 statement · June 30 redeployment

Counterevidence & Limitations

Routine background checks, voluntary vetting, export compliance and a defense contract without the clearance clause do not qualify. A restricted government-only project does not by itself establish coverage of the lab’s general frontier-model staff.

Missing qualifying public evidence does not establish that the event has not happened privately. The assessment concerns the documented test and its observable evidence.

What Would Change Our Assessment

Requirements remain voluntary or limited to a small government project; coverage or implementation does not match the claim.

Evidence satisfying the full test would change the assessment; early qualifying evidence can be considered before the scenario window. Partial proxies are recorded with their scope.

Update History

DateUpdate
2026-11-07Assessment reviewed. Status: Not Yet Testable; confidence: 80%. OpenAI’s February 28 agreement specifies cleared engineers and safety researchers in classified deployment. It does not establish an enforced clearance requirement for everyone working on a general frontier-model program or completion within the scenario’s two-month deadline. Correction: the June access directive was lifted June 30; export restrictions are not evidence of personnel-clearance coverage.
2026-09-06No qualifying enforced clearance requirement covering a general frontier-model program has been verified. Relevant primary evidence and limitations are described above.