Peer Review Is Not Authorization

FERZ Technical Note · v1.0 · July 2026 Peer Review Is Not Authorization examines the frontier-model peer-review proposal set out in the interview published by The Economist on July 23, 2026, and treats it as a recent, high-profile instance of a broader class: release governance by review. The question the note asks is prior to any judgment about the proposal's feasibility: when review concludes, who has authorized what? A review finding characterizes a system, capability, or risk on the basis of testing or analysis; it yields evidence. An authorization evaluates a specific proposed action under the policy and authority state in force; it yields a pre-execution verdict. Measured against the Five Tests Standard (5TS: Stop, Ownership, Replay, Escalation, Provenance), the proposal as publicly described supplies evidence but does not establish any of the five tests. It specifies no mandatory pre-release gate, no final authorization authority, no independently reconstructable determination, no fail-closed handling of unresolved concerns, and no established origin, version, and binding for the findings that would ground a release decision. The note then identifies the conditions under which review findings become bound inputs to authorization rather than substitutes for it: established origin and version; binding to the proposed action and determination; evaluation by a defined authority under an applicable, versioned policy before execution; a verdict from the closed verdict space ALLOW, DENY, or ABSTAIN, with DENY terminal for the requested action and ABSTAIN blocking execution pending authorized human override; a non-bypassable, fail-closed runtime authorization boundary; and preservation of the determination in a tamper-evident authorization artifact from which the verdict can be independently reconstructed. In the vendor-neutral 5TS register, the required proof object is a Proof-Carrying Decision. Peer review identifies the hazard. Authorization determines whether the model may be released or the action may execute, under whose authority, and with what independently reconstructable authorization artifact. Series and corpus. This note is the second entry in the series begun with LLM-as-a-Judge Is Not Authorization (July 2026): an instrument that produces judgments about systems is not thereby an instrument that authorizes actions. It draws on Authority versus Authorization: A Definitional Framework for AI Governance, the Five Tests Standard (5TS), Evidence Projection, the Authorization Boundary Integrity Model, On the Impossibility of Observability-Based Authorization, From Monitoring to Authorization, and The Closed-World Bargain. Publisher: FERZ, Inc. · License: CC BY 4.0 · Canonical URL: ferz.ai/articles/peer-review-is-not-authorization