Enterprise handbook · Reviewed 2 October 2026
Find a control, see the requirements it supports, and follow practical steps to implement and test it. Start with a control or a regulation.
Selecting an instrument does not determine whether it applies to you. Check entity, role, intended use, exclusions, and timing against each source. EU AI Act is one instrument; the EU jurisdiction also includes GDPR and DORA.
169 matching items; 12 shown
H-GOV-01 · Handbook practice
Approve purpose, excluded uses, action classes, and accountable process owner.
Start here: Document the agent's purpose, operating boundary, permitted action classes, excluded uses, and process owner.
Evidence: Signed mandate and system boundary.
H-GOV-02 · Handbook practice
Assess affected people, consequences, jurisdictions, entity roles, and data before choosing autonomy.
Start here: Identify the intended use, affected people, consequential decisions, data, jurisdictions, and organizational roles.
Evidence: Impact register and scoped obligation decisions.
H-GOV-03 · Handbook practice
Assign release, exception, halt, restart, correction, and retirement authority.
Start here: Document permitted decisions and name who can approve release, accept an exception, stop operation, and authorize restart.
Evidence: RACI, delegated limits, and decision records.
H-GOV-04 · Handbook practice
Record specific failure mechanisms, residual gaps, compensating controls, owners, and expiry.
Start here: Identify the intended use, affected people, consequential decisions, data, jurisdictions, and organizational roles.
Evidence: Risk register and expiring exception log.
H-GOV-05 · Handbook practice
Demonstrate operator and reviewer competence, workload capacity, and understanding of role limits.
Start here: Define competence requirements for operators and reviewers, including limitations and escalation rules.
Evidence: Training rubric, planted cases, staffing and burden evidence.
H-DES-01 · Handbook practice
Version model, harness, instructions, tools, permissions, memory, monitors, and fallback.
Start here: Register a stable system identity, owner, purpose, deployment environment, model version, tools, permissions, and lifecycle dates.
Evidence: Deployment manifest and production digest.
H-DES-02 · Handbook practice
Enforce least privilege and resource authorization outside agent-editable policy.
Start here: List allowed actions and resources for each identity and context; issue the minimum permissions needed.
Evidence: Effect-path inventory, grants, and negative outcomes.
H-DES-03 · Handbook practice
Define permitted sources, classifications, rights, freshness, tenant scope, and egress.
Start here: Register data sources, provenance, permitted purpose, access conditions, and an accountable steward.
Evidence: Source register, access decisions, and data-flow records.
H-DES-04 · Handbook practice
Gate admission and reuse; retain provenance, derivative lineage, revocation, and deletion policy.
Start here: Record memory provenance, state versions, dependencies, and who may change or restore them.
Evidence: Admission records, lineage, quarantine and repair evidence.
H-DES-05 · Handbook practice
Specify behavior on monitor, policy, tool, audit, provider, and identity failures.
Start here: Identify dependency failures, partial actions, and the fallback or recovery behavior required for each.
Evidence: Dependency failure matrix and reconciliation design.
H-EVL-01 · Handbook practice
Define authorized utility, prohibited effects, adjudication, and external success criteria.
Start here: Define the intended-use evaluation, representative cases, known failure modes, and decision criteria before running tests.
Evidence: Evaluation protocol and oracle validation.
H-EVL-02 · Handbook practice
Cover normal work, edge cases, missing data, accessibility, and relevant customer stages.
Start here: Define the intended-use evaluation, representative cases, known failure modes, and decision criteria before running tests.
Evidence: Case register, sampling rationale, labels, and uncertainty.
Before relying on a mapping, establish the jurisdiction, entity, intended use, and role in scope. The filters narrow the handbook; they do not perform an applicability assessment. The 45 handbook practices, 12 specifications and 112 reference controls retain separate identities. Each relationship is an editorial interpretation of how a proposed safeguard can contribute; it is not a compliance verdict.
A blank comparison cell means no mapping recorded. It does not establish an exemption. Inspect versions, timing and applicability
Explanation register
These shared topic explanations provide the background for the mappings. They are editorial interpretations of a contribution to a broader objective, not complete control-by-clause compliance assessments. Original explanation identifiers and earlier links remain available.
Connect decisions about use, residual risk and intervention to named people with authority. A charter alone does not demonstrate operation.
Evidence to collect: Approved mandate; accountable owner; release and exception decisions; escalation records.
Identify the deployed system, its dependencies and accountable owner so evidence can be joined to the configuration actually used.
Evidence to collect: System inventory; versioned manifest; dependency and model records; approval linkage.
Assess intended use, affected people, applicable roles and materiality before selecting controls. Record the reason a law applies or does not apply.
Evidence to collect: Impact assessment; jurisdiction and role decision; risk register; residual-risk owner.
Reviewers need competence, time, evidence and the authority to change or stop a consequential action. Measure correction rather than counting approvals.
Evidence to collect: Reviewer test results; workload and response times; held actions; corrections and halt exercises.
Demonstrate that operators, reviewers and affected users understand limitations and their responsibilities. Training attendance is incomplete evidence.
Evidence to collect: Competence rubric; planted cases; accessible instructions; staffing and adoption evidence.
Keep source provenance, permitted use and data quality traceable; additional lawful-basis and rights review is needed when personal data or protected material is involved.
Evidence to collect: Data/source register; legal basis; provenance; permitted use; quality tests.
Limit access and reuse to the authorised purpose and retain personal data only as justified. Test deletion and correction through caches, memory and derived records.
Evidence to collect: Purpose/retention schedule; deletion lineage; access and correction tests; minimisation decisions.
Authenticate the principal and check the actual action and resource before an effect. An authenticated tool call does not establish business authority.
Evidence to collect: Permission matrix; grant/revocation history; denied cross-tenant and expired-authority outcomes.
Prevent untrusted inputs, integrations and changed artifacts from bypassing security boundaries. The specified technical mechanism remains an implementation choice.
Evidence to collect: Threat model; integrity checks; approved artifacts; injection and exfiltration tests.
Trace persistent and derived state, prevent contamination from acquiring authority, and verify that repair reaches descendants. Privacy duties are conditional on personal data.
Evidence to collect: Memory provenance; state lineage; quarantine; corrected descendants; restore tests.
Evaluate utility, relevant failure modes and the deployment decision under declared conditions. A test score does not establish fitness outside those conditions.
Evidence to collect: Evaluation plan; representative cases; configuration; pass/fail decisions; uncertainty and limitations.
Challenge the system and its integrations under credible threats. Specialist attacks are proposed ways to test broader robustness expectations, not named legal requirements.
Evidence to collect: Attack assumptions; coverage; adversarial cases; effect records; residual vulnerabilities.
Separate implementation from challenge, review evidence limitations and give reviewers the standing to change the decision.
Evidence to collect: Independent validation; challenge findings; remediation; audit scope; unresolved limitations.
Detect changes to components, permissions, tasks and suppliers; re-evaluate the affected claims before relying on old approval evidence.
Evidence to collect: Baseline/candidate comparisons; materiality decisions; notices; regression results; new approvals.
Join identity, policy, decisions and actual effects in protected records. Logging is subject to privacy, confidentiality and retention limits; raw reasoning disclosure is not a universal duty.
Evidence to collect: Attributable action log; configuration and approval IDs; integrity/access checks; retention schedule.
Monitor behaviour and outcomes against the tested envelope, including latency, capacity and blind spots. Reopen claims when the environment changes.
Evidence to collect: Signal definitions; thresholds; measured delay/error; drift cases; reassessment decisions.
Contain effects and credentials, preserve evidence, investigate mechanisms, and feed corrections into the next operating decision.
Evidence to collect: Incident classification; containment times; forensic record; root cause; re-entry decision.
Prove safe operation and effect reconciliation under dependency failure. Recovery must preserve permissions and unresolved business obligations.
Evidence to collect: Failure matrix; fallback and restore exercises; duplicate/partial-effect reconciliation.
Inventory suppliers and processing relationships; secure usable documentation, audit rights, change/incident notice and data handling terms.
Evidence to collect: Due diligence; contracts; processor terms; provider evidence; dependency register.
Identify critical dependencies and demonstrate an exit that closes grants, preserves needed evidence, and restores the business process.
Evidence to collect: Concentration assessment; tested exit; data return/deletion; credential withdrawal; custody decisions.
Measure relevant cohort outcomes, burdens and exclusions across routing and intermediate actions as well as the final decision. Statistical disparity alone is not a complete legal conclusion.
Evidence to collect: Covered-decision inventory; cohort/trajectory results; uncertainty; alternative-design analysis.
Explain the actual decision basis and the agent’s role in terms affected people can use. Generated rationalisations are insufficient evidence of the reasons actually applied.
Evidence to collect: Decision factors; validated notice; AI disclosure; intelligibility tests; delivered notice record.
Provide an accessible route to correction or review and investigate recurring harm beyond the person who complained. Rights, exceptions and remedy duties depend on the use and jurisdiction.
Evidence to collect: Access/correction requests; human reconsideration; complaint results; affected-cohort lookback.
Preserve identity, narrower authority, state lineage and containment across agents. Review the whole system rather than assuming individually acceptable components compose safely.
Evidence to collect: Topology; parent/child authority; joint-outcome tests; propagated revocation and containment.
Keep action, exposure and time limits outside agent-editable state, including retries and descendants. Proposed budget mechanisms support risk bounding; the sources do not prescribe universal numeric limits.
Evidence to collect: External limits; consumption ledger; expiry and restart tests; bounded mandate.
Separate organisational certification, system conformity assessment and evidence of actual control effectiveness. This map cannot establish any of those verdicts.
Evidence to collect: Applicable conformity route; certificate scope; independent assessment; exclusions; management-system evidence.
Check external business state before asserting completion and reconcile uncertain effects. This is a proposed implementation of accuracy and evidence disciplines, not a universally mandated oracle design.
Evidence to collect: External postconditions; idempotency; state reconciliation; false-completion cases.
Trace training and retrieved sources and assess protected data or source rights. GPAI model-provider documentation duties do not automatically pass to every agent deployer.
Evidence to collect: Training/source lineage; rights assessment; extraction tests; supplier documentation.
Check measurement grain, labels, oracle quality, independence and uncertainty before relying on a safety or oversight claim. Specialist test methods remain proposed implementations.
Evidence to collect: Protocol; denominator; independent labels; uncertainty; scope and reproducibility notes.
Keep each trigger, recipient, clock and reporting template distinct. Do not apply a single deadline to every AI or ICT incident.
Evidence to collect: Reportability assessment; jurisdiction/authority; clock start; reports and notification receipts.
The dated edition files retain the original topic-based alignments. Each control inherits the references associated with its explanation topics; these files do not certify item-level compliance. For the current selection and suggested procedures, use the filtered implementation export above.
Read Appendix K in the complete PDF or manuscriptPrimary evidence
Reviewed 2 October 2026. Next monthly review due 1 November 2026. Monthly reference updates supplement this dated edition; they do not silently revise its legal map.
R01 · EU · EU
Binding law
Commission consolidated-text explorer, 27 July 2026. Status source.
R02 · GD · EU
Binding law
Official consolidated text: CELEX 02016R0679-20160504; reviewed 2 October 2026. Status source.
R03 · DO · EU
Binding law
Core regulation and official rulebook. Status source.
R04 · FC · UK
Binding rules and guidance
Live Handbook, including June 2026 updates. Status source.
R05 · PR · UK
Supervisory expectations
April 2026 revision. Status source.
R06 · RB · US
Binding regulation
Current CFPB rule text. Status source.
R07 · CA · US
Binding regulation
Final approved 2025 text; effective 2026. Status source.
R08 · CO · US
Enacted law; future duties
Signed 14 May 2026. Status source.
R09 · SR · US
Supervisory guidance
17 April 2026; supersedes SR 11-7 and SR 21-8. Status source, rechecked 2026-10-03.
R10 · IM · Singapore
Voluntary guidance
Version 1.5, 20 May 2026; updated 5 June 2026. Status source.
R11 · SF · Singapore
Voluntary industry white paper
Version 1.0, July 2026. Status source.
R12 · HK · Hong Kong
Regulator guidance
25 August 2026. Status source.
R13 · NI · Global
Voluntary framework
AI RMF 1.0, January 2023. Status source.
R14 · NG · Global
Voluntary framework
NIST AI 600-1, July 2024. Status source.
R15 · IS · Global
Voluntary standard
ISO/IEC 42001:2023. Status source.
R16 · IR · Global
Voluntary guidance standard
ISO/IEC 23894:2023. Status source.
R17 · FS · Global
Consultation draft
10 June 2026 consultation report. Status source.