The Scope Creep Automated reviews of new AI-related working group submissions

Review · August 20, 2026

Government Content Access Control (GCAC) Summary

draft-fx-oauth-government-content-access-control -02View on Datatracker →

Problem Statement

Governments increasingly mandate that online services restrict content based on user age or legal status, but existing mechanisms require personal data disclosure or enable cross-service tracking. This spec defines an OAuth 2.1 profile where a government-operated Authorization Server can answer narrowly scoped eligibility questions (e.g., "is this user allowed to access pornography?") with a boolean response, without revealing PII or allowing relying parties to correlate users across services.

How the Spec Fits In

This is an OAuth 2.1 profile introducing a specialized new role — the Government Content Control Authority (GCCA) — that acts as a constrained Authorization Server. It extends the standard Authorization Code + PKCE flow with two non-standard additions:

  • A scope_results object in the token response (a map of scope → boolean eligibility decision)
  • A two-step HMAC-SHA256 token derivation scheme for issuing RP-scoped pseudonymous site tokens

It references OAuth 2.1 (I-D.ietf-oauth-v2-1), PKCE (RFC 7636), mTLS client auth (RFC 8705), and optionally JAR (RFC 9101). The dependency footprint is modest. It does not build on OIDC, RAR (RFC 9396), or Token Introspection, though these would seem natural companions.

This spec introduces a new role (the GCCA / government-operated AS) and adds a new response parameter (scope_results), but does not extend any single existing RFC in a deep way. The main novelty is the privacy-preserving token derivation and the boolean scope result semantics.

High-Level Summary

A user visits a website (Relying Party) that is legally required to verify the user's eligibility before showing regulated content. The RP redirects the user to a government-operated server (GCCA), which authenticates the user through government mechanisms (national ID, etc.) and evaluates whether that person qualifies under each requested regulatory scope. The GCCA returns an access token along with a scope_results object ({"pornography": true, "social_media": false}). Crucially, the RP never learns who the user is — the access token is an HMAC-derived, site-scoped pseudonymous identifier, so the same person gets different tokens at different RPs, preventing cross-site correlation.

The key claims:

  • The government learns which RPs a user visits (through the authorization flow)
  • RPs learn only eligible/ineligible per scope, not the user's identity or age
  • The cryptographic token derivation prevents collusion between RPs to re-identify users

Analysis and Concerns

Framing vs. actual privacy properties: The spec emphasizes user privacy from RPs, but the government GCCA is necessarily a centralized observer of all access requests. The privacy guarantees are one-directional. A user asking a government server whether they're allowed to view content about gambling, VPNs, or social media creates a detailed behavioral profile. This tension is not addressed.

Scope model is political, not technical: The example scopes — pornography, social_media, gambling, alcohol, firearms, vpn, proxy — vary enormously by jurisdiction and reflect different governments' regulatory priorities. Including vpn and proxy as government-regulatable content categories raises significant civil liberties concerns. The spec does not address how scopes are defined, standardized, or contested across jurisdictions.

The user authentication mechanism is unspecified: The spec says the GCCA "authenticates the user via government-controlled mechanisms" but says nothing about what these mechanisms are. This is the most security-critical part of the entire system, and it's entirely out of scope.

The scope_results parameter is non-standard and underspecified: Returning eligibility decisions as a new top-level parameter in the token response is an unusual extension. There is no discussion of how this interacts with token introspection, JWT access tokens, or downstream token validation. An RP receives a boolean result but the spec doesn't clearly define the semantics of the access token itself — is it an opaque handle the RP presents for future calls, or is the one-time boolean result all that matters?

Token derivation scheme: The HMAC derivation (site_token = HMAC(HMAC(master_secret, national_id), client_id)) is a reasonable pseudonymization scheme, but it assumes a single GCCA holds a master secret covering the entire national population. This is a single point of failure and a high-value target. The spec doesn't address key rotation, compromise recovery, or multi-operator deployments.

Re-verification section is nearly empty: Section 7 says RPs "MUST offer" re-verification but gives no guidance on UX, timing, or how the RP should update its state. This needs significantly more specification.

Writing Quality

The writing is reasonably competent and follows IETF conventions for the most part. RFC 2119 keywords are used correctly. The spec is concise — perhaps too concise in places (re-verification, the user authentication mechanism, the token response semantics). Some sections read more like a design sketch than a finished specification. The abstract and introduction are clear. The token derivation section is the most technically precise part of the document.

Overall: the writing quality is acceptable but several key mechanisms lack the specificity required for interoperable implementation.

Author Background

Francois-Xavier Morin ([email protected], FXCO Ltd.)

  • RFCs published: 0
  • Active Internet-Drafts: 0 (one expired previous version of this same draft)
  • Working group involvement: None

This is a new IETF contributor with no prior RFC publications and no working group affiliations. The draft itself is the only evidence of IETF activity. This is not inherently disqualifying, but it means there is no track record to draw on for assessing the author's familiarity with IETF norms or OAuth ecosystem context.

Overall Assessment

This spec addresses a real and growing policy problem — governments mandating content restrictions — and the privacy-preserving framing is thoughtful. The boolean-response model and RP-scoped token derivation are technically interesting approaches to minimizing data exposure.

However, the spec has significant gaps: the authentication mechanism is unspecified, the scope model is politically loaded without a governance framework, the token response semantics are underspecified, and the centralized government observer model is presented without honest discussion of its privacy implications. The re-verification section needs substantial work.

As an individual draft from a new contributor with no IETF track record, this is unlikely to gain traction without significant revision and broader co-authorship from people with both OAuth expertise and digital rights/policy experience. The working group should also carefully consider whether this type of government-mandated control infrastructure is within OAuth's intended scope.