Govern Regulatory Bodies, Regulations, Regulatory Obligations, Controls, and Evidence

Why Regulatory Inventories Matter

Enterprise AI Governance requires regulatory inventories because AI-related obligations must be translated from external and internal authority into governed enterprise data.

Regulations, standards, contracts, and policies do not govern AI by themselves. They must be identified, decomposed, interpreted, assigned, mapped, controlled, monitored, and evidenced. The enterprise must understand which authorities create obligations, which regulatory instruments contain those obligations, which obligations apply to which AI uses, which controls satisfy those obligations, and which evidence proves that controls operated.

This does not mean that every mapping becomes a separate inventory. It means the enterprise needs clear inventories of the core regulatory Noun Types and a connected Enterprise Model that relates those Noun Instances to AI Use Cases, AI Agents, AI Models, AI Prompts, Applications, Data and Information, Vendors, Locations / Jurisdictions, Risks, Incidents, Controls, and Evidence Records.

The purpose of regulatory inventories is to make regulatory governance operational. The enterprise should move from regulatory text to specific obligations, from obligations to controls, from controls to evidence, and from evidence back to the AI assets and stakeholders the controls protect.

The Core Regulatory Inventory Set

The core regulatory inventory set for Enterprise AI Governance should include Regulatory Agencies (a.k.a. Regulatory Bodies), Regulations, Regulatory Obligations, Controls, Evidence Records, and Locations / Jurisdictions.

These inventories are not all AI-only inventories. In a mature enterprise, they may support many governance domains beyond AI. However, they become essential to Enterprise AI Governance because AI must be governed against applicable authority, location, obligation, control, and proof.

The Regulatory Agencies Inventory identifies the authorities or organizations that issue, interpret, supervise, or enforce obligations. The Regulations Inventory identifies the regulatory instruments, standards, policies, contracts, and other authoritative sources. The Locations / Jurisdictions Inventory identifies the geographic, legal, operational, and data-residency boundaries that affect applicability.

Together, these inventories provide the regulatory backbone for AI governance.

Regulatory Agencies Inventory

The Regulatory Agencies Inventory identifies the authorities (e.g., government bodies, standards bodies, or other organizations) that issue, interpret, supervise, or enforce obligations relevant to enterprise AI use.

A Regulatory Agency may be a supranational authority, national regulator, state or provincial agency, municipal authority, sector regulator, standards body, contractual authority, industry body, certification body, or internal enterprise authority.

For each Regulatory Agency, the enterprise should capture the name, jurisdiction, authority scope, industry or sector scope, related Regulations, enforcement or oversight role, status, owner or steward, review date, and source references.

The value of this inventory is that obligations have origin. If a Regulatory Agency issues a new rule, changes guidance, updates an enforcement posture, or modifies expectations, the enterprise should identify which Regulations, Regulatory Obligations, AI assets, Controls, and Evidence Records may be affected.

Regulations Inventory

The Regulations Inventory identifies the laws, regulations, directives, rules, standards, frameworks, guidance documents, contracts, policies, and other authoritative instruments that may create obligations for enterprise AI use.

A Regulation record should identify the title, type, issuing Regulatory Agency, jurisdiction, effective date, status, version, scope, affected industries, affected stakeholder types, related Regulations, source location, owner or steward, review date, and lifecycle state.

For Enterprise AI Governance, the Regulations Inventory should include AI-specific regulations, but it should not stop there. Privacy laws, employment rules, consumer protection regulations, cybersecurity obligations, sector-specific requirements, records-retention rules, intellectual property obligations, procurement rules, contractual commitments, data residency requirements, and internal policies may all create AI governance obligations.

The Regulations Inventory is the source layer from which Regulatory Obligations are decomposed.

Regulatory Obligations Inventory

The Regulatory Obligations Inventory is the governed inventory of actionable obligations derived from Regulations.

A Regulation may contain many obligations. Some require the enterprise to do something. Some prohibit an activity. Some restrict certain uses. Some require documentation, disclosure, testing, human oversight, logging, monitoring, reporting, retention, review, approval, vendor management, or incident notification.

A Regulatory Obligation record should identify the obligation statement, source Regulation, issuing Regulatory Agency, article, section, clause, paragraph, or source reference, jurisdiction, applicability conditions, affected Noun Types, obligation type, required Controls, Evidence requirements, accountable owner, review cycle, lifecycle state, and approval status.

The Regulatory Obligations Inventory is where legal, regulatory, contractual, and policy text becomes operational governance content. The enterprise should ask which obligations apply to a specific AI Use Case, AI Agent, AI Model, AI Prompt, Application, Data Source, Vendor Product, Location / Jurisdiction, Stakeholder, or AI Output.

Controls Inventory

The Controls Inventory identifies the mechanisms used to satisfy, support, enforce, or test governance obligations.

Controls may be technical, procedural, administrative, contractual, organizational, monitoring-based, human-review-based, or evidence-based. A Control may involve access restriction, approval workflow, human oversight, model evaluation, AI Prompt testing, logging, monitoring, incident escalation, regional blocking, disclosure language, vendor contractual commitments, data-loss-prevention rules, release gates, periodic review, or audit testing.

A Control record should identify the control name, description, control type, owner, supported obligations, affected Noun Types, implementation pattern, operating frequency, testing method, evidence produced, lifecycle state, effectiveness status, review date, and remediation status.

Controls are the operational bridge between obligations and execution. A Regulatory Obligation may state what must be achieved. A Control defines how the enterprise satisfies or supports that obligation.

Evidence Records Inventory

The Evidence Records Inventory identifies the proof that governance decisions, controls, reviews, approvals, tests, monitoring, incidents, remediations, and lifecycle actions occurred.

Evidence Records may include approval records, review notes, AI Use Case assessments, model evaluation results, AI Prompt test results, data approvals, vendor attestations, contract extracts, access reviews, configuration screenshots, release records, logs, telemetry, disclosure records, incident records, remediation records, audit findings, control-test results, and exception approvals.

An Evidence Record should identify the evidence type, description, source, owner, related Control, related Regulatory Obligation, related AI asset or Noun Instance, creation date, effective period, retention requirement, review status, storage location, and lifecycle state.

Evidence must be connected. A file archive is not enough. The enterprise must know what the evidence proves, which Control it supports, which obligation it helps satisfy, which AI assets it concerns, and whether it remains current.

Locations / Jurisdictions Inventory

The Locations / Jurisdictions Inventory identifies the geographic, legal, operational, service, market, cloud, and data-residency boundaries relevant to AI governance.

A Location / Jurisdiction may be a supranational region, country, state, province, county, city, town, municipality, facility, service territory, operating market, cloud region, data residency zone, or other boundary that affects governance.

For each Location / Jurisdiction, the enterprise should capture the name, type, parent location, jurisdictional category, related Regulatory Agencies, related Regulations, owner or steward, status, and review date.

This inventory matters because AI obligations often depend on where AI operates, serves users, processes data, affects stakeholders, produces outputs, or triggers regulated outcomes. The required granularity should be determined by the most demanding applicable obligation.

Relationships Are Not Inventories by Default

Enterprise AI Governance should avoid treating every mapping as an inventory.

The Enterprise Model should normally represent relationships among Noun Instances. For example, a Regulatory Agency issues a Regulation. A Regulation contains a Regulatory Obligation. A Regulatory Obligation applies to an AI Use Case. An AI Agent operates in a Location / Jurisdiction. A Control satisfies a Regulatory Obligation. An Evidence Record proves operation of a Control.

These are relationships. They do not automatically need to become separate inventories.

Creating too many relationship inventories can make governance unnecessarily complex. It can also blur the distinction between Noun Types and relationships. The better pattern is to govern core Noun Instances as inventories and connect those Noun Instances through the Semantic Model.

When Regulatory Relationships Should Be Reified

Some regulatory relationships should be reified when the relationship itself needs governance attributes.

For example, a Regulatory Obligation-to-Control relationship may need control sufficiency, testing frequency, evidence expectations, owner, review date, audit status, and remediation state. In that case, the relationship may be governed as a mapping record.

An AI Use Case-to-Regulatory Obligation relationship may need applicability rationale, legal reviewer, compliance reviewer, approval status, effective date, review date, and evidence. In that case, the relationship may need to be reified.

An AI Agent-to-Location relationship may need approval status, regional restrictions, effective dates, applicable obligations, disclosure requirements, monitoring requirements, incident notification rules, and evidence. In that case, the relationship may need to be governed as a first-class mapping record.

The rule is practical: reify a relationship only when the relationship itself must be owned, reviewed, approved, evidenced, controlled, versioned, or maintained over time.

Core Regulatory Relationship Patterns

Enterprise AI Governance should support several core regulatory relationship patterns.

A Regulatory Agency issues a Regulation. A Regulation contains a Regulatory Obligation. A Regulatory Obligation has applicability conditions. Applicability conditions reference Locations / Jurisdictions, Stakeholders, Data and Information, AI Use Cases, AI Agents, AI Models, AI Prompts, Vendor Products, Applications, and other Noun Types. A Regulatory Obligation is satisfied or supported by Controls. A Control produces or requires Evidence Records. Evidence Records prove governance operation.

In AI governance, important paths include:

AI Agent to Location / Jurisdiction to Regulation to Regulatory Obligation to Control to Evidence Record.

AI Use Case to Stakeholder to Location / Jurisdiction to Regulatory Obligation to Control to Evidence Record.

AI Agent to Data and Information to Data Sensitivity Type to Regulatory Obligation to Control to Evidence Record.

Vendor Product to Contract to Regulatory Obligation to Control to Evidence Record.

AI Prompt to Control to Evidence Record.

AI Incident to Location / Jurisdiction to Notification Obligation to Evidence Record.

These paths allow the enterprise to determine applicability, impact, control coverage, and evidence readiness.

Relationship to AI Inventories

Regulatory inventories must connect to AI-specific inventories.

AI Use Cases provide purpose and context. AI Agents provide operational actors. AI Models provide computational capabilities. AI Prompts provide behavior-shaping instructions. AI Outputs provide generated or derived results. AI Risks identify exposure. AI Incidents identify failures or events. AI Governance Evidence Records prove governance activity.

A Regulatory Obligation may apply to any of these Noun Types depending on the obligation. For example, an obligation may require disclosure for a customer-facing AI Use Case, testing for an AI Model, review of an AI Prompt, monitoring of an AI Agent, retention of an AI Output, or notification after an AI Incident.

Regulatory inventories therefore cannot sit apart from AI inventories. Their value comes from being connected through the Enterprise Model.

Relationship to Technical and Enterprise Inventories

Regulatory inventories must also connect to existing technical and enterprise inventories.

Applications, Platforms, Services, APIs, Workflows, Automations, Data Stores, Vendor Products, Vendor Services, Contracts, Business Processes, Organizations, Roles, Stakeholders, Data and Information, Facilities, and Runtime Environments may all influence regulatory applicability and control implementation.

For example, a Regulatory Obligation may apply because an Application is customer-facing, a Workflow affects employees, a Vendor Product processes sensitive data, a Data Store contains regulated information, a Runtime Environment operates in a restricted region, or a Contract imposes AI-specific commitments.

This reinforces the core Enterprise Model pattern: AI governance is not a disconnected AI registry. It is a connected model of AI assets, enterprise assets, regulatory obligations, controls, and evidence.

Regulatory Change Management

Regulatory inventories must support change management.

Regulatory Agencies may issue new guidance. Regulations may be proposed, enacted, amended, superseded, or retired. Regulatory Obligations may change interpretation. Controls may become insufficient. Evidence requirements may change. Locations / Jurisdictions may become newly relevant. Vendor contracts may change. AI uses may expand into new markets or stakeholder groups.

The enterprise should manage lifecycle states and review cycles for Regulatory Agencies, Regulations, Regulatory Obligations, Controls, Evidence Records, and Locations / Jurisdictions.

When regulatory change occurs, the enterprise should identify which AI Use Cases, AI Agents, AI Models, AI Prompts, Applications, Data Sources, Vendor Products, Locations, Controls, Evidence Records, Risks, and Incidents may be affected.

This is one of the highest-value outcomes of a connected regulatory model.

Governance Questions the Regulatory Inventories Should Answer

For the Regulatory Inventories, governance should answer what exists, who owns it, what is affected, which risks, obligations, controls, evidence, incidents, changes, and gaps require action.