Assurance-Conditioned Continuous Authorization for Autonomous AI Agents
Organizations: Independent Researcher, Canada
Abstract
Autonomous software agents can act through tools, delegate work, and change their models or execution environments while earlier authorizations remain in flight. We present Assurance-Conditioned Continuous Authorization (ACCA), an approach designed to be harness-independent, for maintaining operational authority according to currently justified assurance. It distinguishes evidence validity from contextual applicability, derives assurance claims from eligible evidence, and resolves a multidimensional effective-authority envelope within a principal-declared ceiling and applicable policy. A typed dependency graph supports selective re-evaluation of authorization-relevant changes. An authority epoch changes when effective authority changes; a distinct optional authorization-basis version supports revalidation of action-specific authorization conditions. Protected effect-time enforcement can reject stale actions before a consequential resource change, subject to the available commit mechanism. A compositional Domain Governance Profile expresses domain-specific semantics without granting authority. Controlled scenario tests implemented as a regression suite using a trading profile exercise default-deny rules, evidence-conditioned contraction and recovery, policy divergence, selective non-response, and stale-action rejection at a local gateway before the simulated-broker boundary. The study establishes implementation-level evidence in one domain, not a cross-domain guarantee of harmful-behavior prevention.
Figures & tables
| Control question | Primary object | Representative mechanism |
|---|---|---|
| Request-time action authorization | Whether a particular submitted request, including its material parameters, is permitted under current effective authority and applicable action-specific conditions | A policy decision point (PDP), which may be implemented using a policy engine such as Open Policy Agent (OPA), evaluates the request; a distinct policy enforcement point (PEP) enforces the resulting decision at the protected action or effect path. RBAC, ABAC, UCON, XACML, or AADP-like protocols may supply policy semantics or exchange patterns. |
| Effective-authority continuity | Whether the effective-authority generation under which a consequential action was authorized remains current | ACCA effective authority, authority epoch, and an epoch-bound action authorization. |
| Authorization-basis continuity | Whether a previously issued action authorization remains acceptable after a configured change to supporting action-specific authorization state that leaves effective authority unchanged | Optional authorization-basis version carried or referenced by the action-authorization artifact and checked when required. |
| Effect-time commitment | Whether required version comparisons and resource-changing commit are protected against a concurrent authorization transition | Transaction, conditional write, provider-local fence, protected execution intent, or another explicitly stated commitment protocol. |
| Work | Governed object | Primary mechanism | Boundary relative to ACCA |
|---|---|---|---|
| Earned Authority [ 72 ] | Grant continuity and active effects | Transition envelope; immutable effect ceiling; evidence-gated ascent; formal non-amplification | Establishes the closest ceiling/allocation structure. ACCA adds the explicit validity–applicability–claims–applicable-policy derivation and DGP operationalization. |
| CapScope / AAE [ 4 , 24 ] | Capability scope and authorization representation | External typed capabilities; machine-evaluable mandate and constraints | Can represent or constrain authority; ACCA derives why a particular authority state remains justified. |
| AADP [ 53 ] | Individual consequential actions | Stateful per-action decisions using budgets, approvals, reservations, obligations, and kill switches | Supports per-action authorization using mutable decision state; ACCA additionally derives and maintains actor-level effective authority from currently applicable assurance evidence and policy. |
| MI9 / ACS / OpenAgentFlow [ 64 , 30 , 8 ] | Runtime agent behavior, policy decisions, and pending actions | Continuous permission adaptation; deterministic policy decisions; shared pre-execution enforcement | Provide runtime governance and enforcement substrates around ACCA ’s assurance-conditioned authority state. |
| MasuGate / TCT [ 50 , 16 ] | Committed effects and derivation dependencies | Policy-state or cognitive serializability; commit-time validation and recertification | Protect effect-time correctness; ACCA determines the assurance-grounded authority supplied to authorization. |
| Revocation coherence / root-scoped quiescence [ 48 , 74 ] | Distributed revocation and residual effect paths | Coherence bounds; root cuts, sink fences, carrier closure, and cutset evidence | Addresses revocation propagation and residual-effect closure under stated system and enforcement assumptions; complementary to effective-authority derivation. |
| Contract | Mechanism | Resulting guarantee |
|---|---|---|
| Request-time | Policy and epoch are evaluated when the request is admitted. | The decision is current at admission, but delayed execution can outlive it. |
| Effect-adjacent mediated | A credential-holding adapter revalidates immediately before an external call. | Narrows but does not eliminate the race before non-cooperating provider acceptance. |
| Protected intent | Epoch validation and creation or release of a controlled execution-intent record are atomic; unreleased intents can be fenced or cancelled. | Strong through the controlled intent boundary; later provider acceptance remains subject to the provider’s semantics. |
| Provider-fenced | The provider checks a required authoritative fencing value or authorization-state generation and rejects operations stale under its enforced currentness rule. | Protects through provider acceptance only under the provider’s synchronization and acceptance semantics. |
| Transactional | Epoch comparison and governed-resource mutation occur in one serializable transaction, conditional write, or equivalent critical section. | Prevents a stale local commitment when the required authorization-state comparison and mutation share a protected consistency or serialization boundary. |
| Case | Property | Observed result |
|---|---|---|
| S1 | Declared authority/default deny | Authorized request reaches enforcement boundary; out-of-scope and undeclared consequential actions are denied upstream. |
| S2 | Core-origin authority-material change | Model substitution alters the applicable assurance basis and produces policy-dependent authority contraction. |
| S4-U | Trading-origin applicability change | Valid evidence becomes non-applicable in a novel regime; assurance and authority contract and later recover under new epochs. |
| S7 | Policy specificity | Identical assurance facts produce different effective authority under different applicable policies. |
| S9 | Authorization-relevance selectivity | Unrelated context change produces no assurance, authority, or epoch transition. |
| S10 | Local stale-authorization rejection | Authorization issued under a prior epoch is rejected before consequential execution after an authority-material change. |
| Operation | Mean ( s) | Run-mean range ( s) | P95 range ( s) |
|---|---|---|---|
| Domain-profile authorization | 0.468 | 0.460–0.476 | 0.500–0.600 |
| Authority epoch check | 0.080 | 0.079–0.081 | 0.100–0.100 |
| ACCA non-authorization-relevant context change | 1.024 | 0.997–1.041 | 1.100–1.100 |
| ACCA authority-material change + authority recalculation | 3.623 | 3.553–3.724 | 3.900–4.200 |
AI Agent Governance
AI agent governance studies how to authorize, constrain, monitor, and hold autonomous and multi-agent systems accountable.
See all 468 papers