Consent Record
The permission object that ties a subject, asset, purpose, and actor together.
Schema →
Docs ยท OConsent
Build consent-aware AI systems.
Learn how to model, verify, revoke, and audit machine-readable consent across datasets, agents, applications, and AI pipelines.
What OConsent is
OConsent is not a consent banner. It is an implementation layer for consent-aware AI systems. Applications, agents, datasets, and pipelines can check whether a specific actor is allowed to use a specific asset for a specific purpose.
The minimum developer flow.
Five steps take you from a consent record to an enforced, audited decision.
- Create or load a consent record.
- Verify permission before use.
- Enforce the allow or deny decision.
- Log an audit event.
- Respect revocation on the next check.
The consent model.
Subject
The person, creator, customer, or organization granting consent.
Asset
The dataset, file, conversation, memory, or derived resource covered.
Purpose
The intended use, such as llm_training or agent_memory.
Actor
The app, model, agent, or pipeline requesting use.
Scope
Allowed and excluded operations, retention, and other constraints.
Verification Request
The point-of-use question asked before access.
Schema →Verification Response
The allow or deny decision, with a reason code.
Schema →Revocation Event
A record that a permission is no longer valid for future use.
Schema →Audit Event
Evidence that a decision was checked and where it was enforced.
Schema →Four conceptual operations.
The interface the SDK and verification service are built around.
issue()
Create a machine-readable consent record.
verify()
Check whether a specific use is allowed at the point of use.
revoke()
Invalidate a consent record so future verification fails.
audit()
Record or export evidence of a decision.
Draft reference model.
The consent record, verification request and response, revocation event, and audit event are defined in the draft specification. It is a reference model, not a finalized standard, and is designed to map to LLMConsent-style consent records.
What is available today.
Status
The public site currently documents the OConsent reference model and implementation direction. The hosted verification service and published SDK package are in progress. The current public implementation lives in the GitHub organization.
LLMConsent compatibility
LLMConsent defines what AI consent should mean. OConsent makes it enforceable in software.
OConsent is designed to support open AI consent standards and is built for LLMConsent-style consent records. OConsent does not own or control the LLMConsent standard. It is implementation infrastructure that aims to stay compatible with emerging LLMConsent-style consent flows.
Legacy protocol and cryptographic anchoring notes
Historical
Earlier OConsent work explored blockchain-based anchoring and smart-contract verification. The current positioning treats these as optional proof infrastructure, not the primary product story.