---
slug: informed-consent
title: Informed Consent
state: super-draft
id: null
repo: null
proposed_by: ben.stull@wiggleverse.org
proposed_at: '2026-05-29'
graduated_at: null
graduated_by: null
owners: []
arbiters: []
tags: []
---

RFC-0001 (Human) names Informed Consent as one of the entities that every subsequent RFC depends on, and grounds it directly: a Human holds epistemic authority over their experience, body, needs, and intentions, and is the inherent expert on their own meaning. Informed Consent is what turns that authority into a usable constraint. If a Human is the authority on themselves, then any system that models, stores, acts on, or speaks about a Human must have a defensible account of when it is permitted to do so — and that account is consent. This RFC establishes that account as a first-class entity in OHM rather than leaving each downstream RFC to reinvent it.

The need is concrete. Identity asks what may be recorded about a Human; Relationality asks what may be shared between Humans; Agency asks what may be done on a Human's behalf; Wise Speech asks what may be said about or to a Human. Each of these is, underneath, a consent question — and without a shared definition each RFC would answer it differently, leaving the framework with no coherent position on the single question that most determines whether OHM treats Humans as subjects or as data. Consent is also the entity most likely to be flattened into a checkbox if it is not defined carefully, so it deserves the same substantial treatment RFC-0001 gave to the definition of Human.

This RFC will define what consent *is* in OHM and what any valid representation of it must account for, without enumerating every mechanism a deployment might use to obtain it. The intended scope includes: the necessary conditions for consent to be informed (disclosure adequate to the decision, comprehension, and the absence of coercion); the requirement that consent be specific to a purpose rather than blanket, and time-bound rather than perpetual; revocability as a property of the Human and not a favor granted by the system, including what the model owes when consent is withdrawn after data has already been acted upon; the distinction between consent, assent, and authorization-by-proxy for Humans who cannot consent for themselves (children, the incapacitated, the deceased), which RFC-0001's Personhood position directly conditions; and the asymmetry of power between a Human and a system, which means consent obtained under a meaningful imbalance is not freely given. It will also take a position on what consent is *not* — not a one-time gate, not transferable, not inferable from silence or continued use — so that downstream RFCs and deployments inherit a definition that resists the usual erosions.

With Informed Consent defined as a stable entity, the RFCs that follow can be proposed and evaluated against a shared answer to "is this permitted, and by whom," rather than re-litigating the permission question on each one's own terms — the same coherence argument RFC-0001 makes for the definition of Human, applied to the constraint that protects them.