GymGov.org presented the Certification Governance Architecture (CGA) to attendees of the FSMB Annual Meeting 2026. This page collects the key materials and the central argument for medical licensing leaders: the problem isn't the browser — it's the failure to audit cognition.
State medical boards govern physician licensing through high-stakes examinations. AI has broken the assumptions those exams were built on.
Medical licensing boards have the highest stakes, the clearest governance mandate, and the most to gain from a verifiable reasoning standard.
The CGS allows a state medical board to operate a CGS-Certified Arena — an independent adversarial certification environment that accepts Experience Credentials from any CGS-compliant Learning Gym, regardless of which AI platform trained the candidate.
Under the Epistemic Engineering Economy (EEE), a medical board can author and register Reasoning Challenges — high-fidelity clinical reasoning scenarios — into a global artifact registry. Every time a candidate trains or is certified using that challenge, the board receives automated attribution and royalty.
The CGS is an open specification. A candidate who trained in California can present their Experience Credential to an Arena administered in New York. No single platform controls the credential, and no medical board needs to vet the training environment — only the Arena result.
Documents prepared for the FSMB Annual Meeting 2026 and the broader medical licensing community.
The Federation of State Medical Boards governs physician licensure for the United States. In 2026, the central challenge is no longer who has access to medical knowledge — it is how to verify that a physician candidate can reason with that knowledge under pressure, without AI assistance. Locked browsers and static question banks were designed for a pre-AI world. The Certification Governance Architecture replaces perimeter security with cognitive auditing: the candidate is placed inside an Adversarial Arena that generates dynamic interrogations from a truth-graph anchored to clinical ontologies. The candidate cannot memorize the pivot — it is derived in real time from their own prior reasoning. If they can defend it, they pass. If they cannot, it does not matter what their training telemetry showed.
If the CGA is relevant to your work in medical licensing, state board governance, or physician certification, the next step is a direct conversation or the validation form.
