Govern, Map, Measure, Manage — for every AI system you run.

The AI Risk Management Framework organizes AI risk into four functions — Govern, Map, Measure, Manage — and is voluntary guidance, not a standard anyone certifies you against. Govern runs across the other three rather than preceding them. Map establishes context and what the system is actually for; Measure tests it; Manage decides what to do about what you found. A generative-AI profile published in 2024 applies the same four functions to models that generate content.

The hard part is not the framework. It is that Map assumes you know which AI systems you run, and most organizations do not — models arrive inside products, embedded in a vendor's feature release, or built by a team that did not think of it as an AI project. A framework applied to an inventory that is missing half its entries measures the wrong half.

Who it applies to

Any organization that designs, develops, deploys or uses AI systems — which now includes organizations that would not describe themselves as doing anything with AI. Because the framework is voluntary, nobody serves you with it. It arrives through customer due diligence, insurer questionnaires, and internal policy written by people who need a defensible structure to point at.

Voluntary is not the same as optional

The AI RMF has no enforcement behind it, and it is still the structure your obligations get expressed in. It crosswalks cleanly onto ISO 42001, which is certifiable, and onto the EU AI Act, which is law — three regimes that ask overlapping questions about the same systems in three vocabularies. Answering them from three separate inventories guarantees they disagree. One inventory of AI systems, with purpose, owner, risk classification and evidence held per system, answers all three and stays consistent because there is nothing to reconcile.

What it asks, in operating terms

Read as an operating requirement rather than voluntary guidance, the four functions reduce to four standing asks — each answerable from a live inventory rather than a survey.

What you'd actually look at

In the dashboard, every figure opens on click to the system, the evidence and the person behind it. This excerpt is what an AI inventory is made of:

Where teams usually start

With a demo walked through by TruSecure — the AI systems you already run pulled into one inventory, a single system opened to its assessment and its owner, the same record answering AI RMF, ISO 42001 and the AI Act. Onboarding then produces the first baseline; the subscription keeps it current. Packaging is scoped in the conversation, not a price list.

TruSecure helps operationalize requirements and prepare evidence. Legal interpretation should be validated by qualified counsel.