AID-R-003.001: Post-Incident Identity & Session Re-establishment

Version 1.20261015 · Actionable sub-technique

Create new trusted interactive sessions only after incident-scoped credential and session eviction, fresh policy-required reauthentication, and a new identity-provider readback.

Pillars: infra, app

Lifecycle: improvement

Scope boundary

Owns creation and independent verification of new trusted interactive sessions after required eviction and reauthentication. It does not revoke prior credentials, tokens, grants, or sessions and does not restore conversational state or privileges.

AID-E-001: AID-R-003.001 creates a new trusted interactive session; AID-E-001 evicts the compromised credential, token, session, principal, or grant population. This recovery control consumes verified AID-E-001 results and cannot treat successful new authentication as proof that old authentication and authorization objects were revoked.

Implementation guidance

AID-R-003.001-G001

Require MFA or step-up authentication for every applicable restored interactive session under the authority-signed recovery policy.
Purpose and boundary

Use this identity-provider integration reference architecture to verify that every applicable restored interactive session was created only after the policy-required MFA or step-up event and is bound to a fresh post-reset authentication context. It does not rotate workload credentials or prove authorization for tools.

Sign the session-reset and authentication policy

The identity recovery authority signs the incident, tenant and application scope, exact affected-principal and revoked-session population, reset cutoff, allowed factor and assurance classes, required token claims, IdP tenant/policy versions, exempt non-interactive identity classes, evidence freshness, and exception policy. Factor strength and reauthentication windows come only from that versioned policy.

# File: recovery/contracts/session-step-up-policy.yaml
schema_version: aidefend.r003.session-step-up-policy.v1
required_bindings:
  - system_or_tenant
  - incident_id
  - recovery_release_id
  - reset_cutoff_at
  - affected_principal_manifest_digest
  - revoked_session_manifest_digest
  - idp_policy_digest
  - allowed_factor_classes
  - required_acr_values
  - required_amr_rules
  - exception_policy_digest
receipt_required_fields:
  - expected_principal_count
  - observed_principal_count
  - expected_new_session_count
  - verified_new_session_count
  - nonconforming_session_ids
  - unmeasured_principal_ids
  - result
Enforce and retain evidence

Configure MFA/step-up at the IdP or authorization server, not by trusting an application header. Consume and independently verify AID-E-001's exact closed eviction receipt for the pre-cutoff session population, then require a fresh authentication transaction, validate issuer, audience, signature, nonce, auth_time, acr, and amr, then issue a new session under a recovery release identifier. The owner retains signed policy exports, revocation receipts, authentication events, token-claim digests, exceptions, counts, and a detached-signed result in retention-locked evidence; do not retain raw factors or reusable tokens.

Independent readback

A read-only verifier outside the IdP-policy owner trust domain derives the expected principals from the signed population, reads revocation and authentication events directly from the IdP, verifies a policy-defined complete new-session population, and performs positive/negative login tests for every applicable assurance branch. Dependencies are the signed affected-principal/session manifests, IdP trust roots, enforceable policy, trusted time, complete authentication telemetry, and evidence storage. Revocation success is input eligibility, never this guidance's PASS.

Action: authorize restored interactive sessions only from independently verified post-reset IdP evidence; a successful application login alone is not MFA proof.

Open-source tools

Commercial / hosted tools

Threat mappings

MITRE ATLAS

MAESTRO

OWASP LLM Top 10 2026

OWASP ML Top 10 2023

OWASP Top 10 for Agentic Applications 2026

NIST Adversarial Machine Learning 2025

Cisco Integrated AI Security and Safety Framework

Google Secure AI Framework 2.0 - Risks

Databricks AI Security Framework 3.0

AID-R-003.001: Post-Incident Identity & Session Re-establishment | AIDEFEND

Loading interactive view…

A Practical Knowledge Base for AI Security Defenses

Version