AI Agent Authorization Infrastructure

Proof of permission for the agent economy.

Tracking verifiable delegation, scoped authority and authorization infrastructure for AI agents.

Explore the Index
The thesis

Identity is not authority.

An agent can be authenticated and still lack permission for a specific action. Delegation Proof tracks the emerging technical layer that connects human or organizational intent to scoped, constrained and auditable agent authority.

Who delegated?

Human, organization or upstream agent.

To which agent?

A uniquely identifiable autonomous actor.

Under what scope?

Permitted actions, limits and constraints.

For how long?

Expiry, revocation and audit lifecycle.

Tracked drafts · Work in progress

Signals are converging.

These are public Internet-Drafts, not IETF standards or endorsements. We track them because multiple independent proposals are addressing delegation, authorization evidence and constrained authority for agents.

Delegation Receipt Protocol for AI Agent AuthorizationActive I-D

Proposes delegation receipts for AI agent authorization.

View Datatracker ↗
AI Agent Authentication and AuthorizationActive I-D

Addresses authentication, authorization, auditing and delegation using existing standards.

View Datatracker ↗
Agent Identity Protocol: Verifiable DelegationActive I-D

Explores cryptographic artifacts binding identity, authorization, scope constraints and provenance.

View Datatracker ↗
Delegation Proof Index · Preview

Mapping the authorization layer.

A living taxonomy of the primitives emerging around agent authority.

PrimitiveQuestionSignal
Agent identityWho is acting?Active
Delegated authorityWho allowed it?Emerging
Scoped authorizationWhat may it do?Active
Delegation receiptsCan authority be evidenced?Early
RevocationCan authority be withdrawn?Emerging
Founding member access

Building in this layer?

Join the early index of builders, researchers and companies working on agent identity, authorization, delegation and verifiable authority.