A token claim used to identify the actor performing an action on behalf of a principal. In delegated AI and service flows, it separates the human principal from the runtime actor so downstream systems can enforce policy and generate accurate audit records.
What the act claim does in delegated flows
An act claim identifies the runtime actor that is performing an action on behalf of another principal. In delegated and agent-mediated flows, that separation is what lets systems distinguish who initiated the request from which identity is executing it, so policy decisions and downstream records stay attributable.
That distinction matters because delegation is not the same as impersonation. A well-formed act claim preserves the actor chain, so a service can recognise that authority was conveyed rather than replaced, and can apply different rules to the acting entity, the subject, and the transaction.
For implementation guidance on the surrounding flow, OAuth Token Exchange and Delegation Guide explains how actor and subject tokens fit into RFC 8693 token exchange.
Why act claims matter for policy enforcement
Systems that only inspect the subject of a token can miss the actual actor behind a delegated action. The act claim gives downstream policy engines a second identity axis, which is useful when approval, scope reduction, consent, or step-up controls depend on who is executing the operation, not just who benefited from it.
That is especially important in agentic and service-to-service flows, where one runtime may act for many principals over time. If the act claim is absent or ignored, logs, authorisation decisions and audit trails can collapse different actors into one misleading identity, making it harder to apply least privilege and harder to prove who did what.
NHIMG’s OWASP Agentic Top 10 ASI03 Guide shows how task-scoped delegated tokens and per-action checks reduce identity and privilege abuse in agent flows.
How act claims relate to delegation chains
An act claim often appears with other token-exchange constructs that describe a chain of authority. In a simple on-behalf-of exchange, it may identify the immediate actor; in longer chains, it helps preserve the sequence of delegation so each hop can be evaluated, audited, or constrained independently.
That chain awareness is what makes attenuated authority possible. Instead of passing an unchanged credential through every intermediary, the system can express who acted, for whom, and under what delegation context. The practical benefit is cleaner boundary enforcement, better traceability, and less risk of one component silently inheriting more power than intended.
NHIMG’s Delegation Chains and Attenuated Authority for AI Agents explores actor claims, transaction tokens and delegation depth.
Audit, attribution and trust boundaries
The act claim is also an audit object. When it is recorded consistently, logs can distinguish the human principal, the acting service or agent, and the exact delegated transaction. That separation is essential for incident review, non-repudiation, and investigations that need to reconstruct how authority moved through a workflow.
It also helps preserve trust boundaries between components. A receiving system should treat the act claim as part of the security context, not as decorative metadata. If the claim is not validated, is rewritten in transit, or is not propagated at all, the platform can lose the very attribution signal it needs to enforce policy correctly.
RFC 8693: OAuth 2.0 Token Exchange defines the token exchange model that gives the act claim its standard delegation context.
Risk and Threat Considerations
Act claims create value only when they are trusted, preserved and interpreted correctly. If they are missing, forged, flattened into a single identity, or ignored by downstream services, delegated actions can become indistinguishable from direct actions, which weakens authorisation, attribution and incident reconstruction.
Failure mechanism: An attacker or misconfigured intermediary can abuse token exchange, strip actor context, or rely on a receiver that does not validate the delegation chain, causing the system to attribute an action to the wrong identity or to over-trust a delegated request.
Impact: The result can be privilege abuse, incorrect policy decisions, misleading audit trails, and slower post-incident analysis because defenders cannot reliably tell which actor executed the action and under what authority.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP API Security Top 10 addresses the attack and risk surface, while NIST SP 800-53 Rev 5, NIST SP 800-63 and NIST Zero Trust (SP 800-207) set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST SP 800-53 Rev 5 | IA-9 — Identification and Authentication (Service and Other Systems) | Covers system-to-system authentication where delegated actors exchange tokens. |
| AU-3 — Content of Audit Records | Audit records need actor and subject context to preserve attribution in delegated flows. | |
| Recommendation — Use IA-9 to authenticate the acting service or agent before accepting delegated authority. Include actor and delegation details in audit records so downstream review can separate principals from runtime actors. | ||
| OWASP API Security Top 10 | API2 — Broken Authentication | Token exchange and actor context affect how APIs authenticate delegated requests. |
| Recommendation — Validate delegation-aware authentication so APIs do not treat every bearer token as direct user authority. | ||
| NIST SP 800-63 | Digital Identity Guidelines | Supports assurance and federation concepts behind delegated identity assertions. |
| Recommendation — Align federated identity handling with assurance expectations before trusting delegated assertions. | ||
| NIST Zero Trust (SP 800-207) | Zero Trust Architecture | Zero trust requires continuous evaluation of actor context and least-privilege access decisions. |
| Recommendation — Continuously verify the acting entity and restrict delegated access to the minimum necessary authority. | ||
Practitioner Guidance
Why practitioners should care: Treat the act claim as a control point, not just a token field. If your platform uses delegation, make sure the runtime actor, the principal, and the delegation path are all visible to the components that enforce policy and produce logs.
Common misunderstanding: A valid token does not automatically mean a valid delegation story. The receiving service still needs to understand whether the token represents direct authority, delegated authority, or an impersonation-style flow, because those cases do not carry the same security meaning.
Practitioner takeaway: Keep actor context intact end to end, and verify that every service that consumes delegated tokens can use the act claim in both authorisation decisions and audit records.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Free weekly newsletter
Subscribe to the NHI & AI Identity Journal
The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.
Bonus 33% off our NHI Course when you subscribe.
Reviewed and updated by the NHIMG editorial team on October 30, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org