Four functions, published by NIST in January 2023 as AI 100-1: Govern, Map, Measure and Manage. It sorts AI risk work by what an organisation actually does about that risk rather than by what it happens to build, and it is voluntary from its first page through to its very last one.
NIST AI RMF 1.0 · VOLUNTARY · SELF-ATTESTED
Aligned with the NIST AI Risk Management Framework.
Every audit we run is built on its four functions - and nobody on earth certifies it, including us.
Govern
Map
Measure
Manage
NIST AI 100-1
Our promise
“A framework is words. The profile is evidence.”
Every finding is written against a clause of the instrument itself — defensible line by line, to anyone who asks. The fee is fixed at $12,500, and nothing is charged until you approve it.
- Each additional system
- $3,000
- Re-audit, same scope
- $8,000
- Renewal, every twelve months
- $10,500 locked
This costs more than the estate ladder, and it should. The ladder is a private assessment written for you. A framework audit produces a published summary iDharma maintains for twelve months - a notice template where the law requires one, a 60-day expiry warning, and a quarterly check that the summary is still live and still linked.
Request this auditThe case file
The AI RMF, in three chapters
Most teams assume there is a certificate waiting at the end of it, because somebody offered them one. There is not: neither NIST nor CAISI certifies anybody, at any tier, at any price whatsoever. A voluntary framework has no compliance state for anyone to be in.
Our audit is built on the four functions and says so in public, category by category. We map each finding to the one that it answers, record an explicit gap wherever the evidence is missing, and keep the entire working record in a hash-chained trail you can check for yourself later.
The four functions
Four functions. You run all of them.
The Core sorts work by what you do about risk, not by what you happen to build.
- Govern - the culture the other three run inside
- Policies, named owners, workforce, third-party risk
- Map - context established before anything is tested
- Purpose, categorisation, capabilities, impacts on people
- Measure - evidence, or an explicit gap. Never a guess
- Methods, trustworthy characteristics, tracking, efficacy
- Manage - what actually gets done about what you found
- Prioritised treatment, vendor risk, monitoring, records
Know who is speaking
The framework is theirs. The claim is ours.
Publishes the framework itself
The AI RMF 1.0 came out of NIST's Information Technology Laboratory in January 2023 as NIST AI 100-1, with the Playbook and the AI 600-1 Generative AI Profile beside it. It is voluntary by design, it is officially under revision, and it confers no status at all on anyone who follows it.
Runs the testing, not the framework
The Center for AI Standards and Innovation - the U.S. AI Safety Institute until it was renamed in June 2025 - is industry's primary federal contact for AI testing. It publishes the AI 800 series and its model evaluations. It does not own the AI RMF, which is why we cite the two of them separately every time.
Neither of them certifies anyone
There is no “NIST certified” and no “CAISI certified” programme, in any tier, at any price. A voluntary framework has no compliance state for anyone to be in. A vendor selling you one is selling something that does not exist, and we would rather you heard that here than from them later.
What a vendor will tell you
“We’re NIST certified — and we’ll certify you.”
What is actually true
No such programme exists. Alignment is self-attested.
Including ours — which is why this page shows its working.
- Who it is for
- AI product teams
- Boards & risk committees
- Banking & insurance
- Federal contractors
- Procurement & vendor risk
What we say & won’t say
02
We won’t say “NIST certified”, “CAISI certified”, “CAISI compliant”, or “compliant with the AI RMF”. None of them exists.
03
Reports state the framework version they were assessed against. A voluntary framework has no compliance state to be in.
04
When the AI RMF revision publishes, this page and our mapping get re-verified against it. Our alignment is self-attested, and the working record is verifiable.
The 60-second check
Three questions. Then you’ll know.
No email, no signup. A starting point, not a determination.
Opens in a panel. Close it any time - your answers are kept.
NIST AI RMF 1.0
Your function check
The calendar
Four moments, and one has not arrived.
Three of these are published and one is still ahead - so they cannot be cited as one body of guidance, and the version you read is part of the claim.
-
AI RMF 1.0
January 2023NIST AI 100-1 published, with the Playbook beside it. Still the operative version, and the one we cite.
-
GenAI Profile
July 2024NIST AI 600-1 adds the cross-sectoral profile for generative AI, and it is where our agentic tests begin.
-
CAISI
June 2025The AI Safety Institute becomes the Center for AI Standards and Innovation. Same body, broader remit.
-
AI RMF 2.0
Under revisionA revision to AI 100-1 is in progress with no published date. We re-verify this page when it lands.
The trap
Two more documents get cited as if they were settled: NIST AI 800-1 and AI 800-2 are public drafts. We read both and use them, and we flag them as drafts every time — because a draft quoted as guidance is how a careful page starts overclaiming.
The mapping
What the function asks, what we do
20 categories, and the audit step that answers each one. Paired, so every claim on this page can be checked against AI 100-1 beside it.
- Policies and processes Govern 1 - risk management is a culture, not a file
- The audit opens on your AI policies: what exists, what is followed, and where the two differ.
- Accountability structures Govern 2 - somebody owns each system, by name
- Ownership traced per system to a named role, with the gaps recorded as findings rather than assumed.
- Workforce capability Govern 3 - the people running it can run it
- Whether the teams operating each system hold the training the policy says they do, evidenced.
- Engagement with actors Govern 5 - the people affected have a route in
- Feedback and complaint routes for affected users tested end to end, not read off a policy page.
- Third-party AI risk Govern 6 - your supply chain is your risk
- Model and vendor dependencies inventoried, with what each contract actually discharges written down.
- A risk-aware culture Govern 4 - considered, and communicated
- How an AI risk actually gets raised and escalated, traced through your incidents rather than your policy.
- Context established Map 1 - what the system is for, and for whom
- Intended purpose, users and operating context agreed in writing before any testing begins.
- System categorised Map 2 - the task, the method, the deployment
- Each system categorised by use and exposure, including its EU AI Act risk class where relevant.
- Capabilities understood Map 3 - benchmarked, not assumed
- What the system can and cannot do, checked against the benchmarks its own documentation cites.
- Risks mapped Map 4 - across components, bought-in ones included
- Risk mapped over the whole chain - your code, your data, and every third-party element inside it.
- Impacts characterised Map 5 - to individuals, groups and society
- Who is affected and how badly, ranked by consequence to those people rather than by ease of fixing.
- Methods and metrics Measure 1 - appropriate to the system, and applied
- The test method stated per finding, so a reviewer can repeat it rather than take our word for it.
- Trustworthy characteristics Measure 2 - valid, safe, secure, fair, explainable
- Each characteristic evaluated where evidence exists - and an explicit gap recorded where it does not.
- Risk tracking Measure 3 - mechanisms in place and running
- Whether you can detect a new risk between audits, and what happens when one of your controls fires.
- Measurement efficacy Measure 4 - the measuring is measured too
- Findings fed back against your own metrics, so a control that looks green is checked for why it does.
- Risks prioritised Manage 1 - acted on in order of consequence
- Every report closes on a roadmap ranked Immediate, Short-term and Medium-term, with owners named.
- Benefit maximised Manage 2 - strategies that keep the value
- Where a control would cost more than the risk it treats, the report says so instead of piling on.
- Third-party risk managed Manage 3 - the bought-in parts, actively
- Vendor risk carried into the roadmap with a re-check date, not closed at contract signature.
- Documented and monitored Manage 4 - treatment, response and communication
- Continuous Monitoring re-checks the findings monthly: report currency, control health, regulatory watch.
- Our own working record Govern 1, turned back on the auditor
- Every audit step written to an append-only, hash-chained trail that you can verify after the fact.
The engagement
The four functions, run in order
Their sequence, not one we invented to fit a rail.
-
Govern
Policies, ownership and third-party dependencies, examined first.
-
Map
Every AI system inventoried, categorised by use and by exposure.
-
Measure
Evidence-based findings only, each with a severity and a source.
-
Manage
A prioritised roadmap, then a monthly re-check - dated each time.
Struck in your favour
Why teams choose iDharma to read the framework with them
Genuinely independent
We build, resell and operate no AI systems of our own, and take no fee tied to what we find.
Mapped, not asserted
Every finding names the AI RMF category it answers, so a reviewer can check it against 100-1.
We say what we are not
No certificate, no seal, no compliance claim - the framework has no such thing to give anyone.
Our own record is open
The trail behind your report is hash-chained and verifiable: the bar we hold you to, held to us.
Four marks, struck on every report.
Deliverables
What you get
Concrete artefacts, each with a name and a format - you know what lands before you buy.
AI RMF alignment profile
The full assessment in one document: every AI system placed against Govern, Map, Measure and Manage, each category either answered or recorded as an explicit gap, the trustworthy characteristics evaluated wherever evidence exists, and the whole of it written in plain language rather than the framework's.
AI system register
Every system, its purpose, its owner and its status, in a workbook your own team can keep current after we hand it over.
Function-by-function map
Each AI RMF category set against what you hold today, so a thin function shows as thin rather than averaging into one score.
Governance gap list
Policies, ownership and vendor dependencies read against Govern 1 to 6, each gap with a severity and the evidence for it.
Scope memo
Which systems were in scope, which were not, and why each call was made - the Map function's written record, and yours to keep.
Remediation roadmap
Where the gaps sit and what to do first, ranked Immediate, Short-term and Medium-term by consequence to the people affected.
Verifiable audit trail
The append-only record of every audit step, hash-chained and checkable, so our Govern claim can be tested rather than believed.
Format & fee
Real numbers, upfront.
- Scope
- All four functions, every category
- Input
- Your systems and what exists on them
- Re-read
- Annually, or on material change - $10,500 against your known baseline
The Core fixed the scope, not us, so the fee is flat - nothing to meter, and nothing charged until you approve it.
Request this assessment
NIST AI RMF · Alignment profile
$12,500 flat
- Profile across all four functions
- System register, purpose by purpose
- Function-by-function gap map
- Verifiable audit trail included
Show your hand
Four things you have to be able to produce
A voluntary framework is still evidenced or not evidenced. Each of these is either in your hand on the day a board asks, or it is not.
The estate,
inventoried
A list of every AI system you build, buy or embed, with an owner against each. Map 1 and Map 2 both start here, and nothing after them is worth more than this is.
The purpose,
written down
What each system is for, who it is used on, and what it is explicitly not for. An intended purpose held only in somebody's head cannot be governed or measured at all.
The testing,
dated
Evidence that the trustworthy characteristics were actually tested, with the method and the date on each result. Measure 1 to 4 are a practice, not a policy.
The record,
underneath
The logs, the sign-offs and the oversight decisions the report asserts, each of them dated. The record is the evidence; the report is only ever the claim about it.
Four cards, and the date on each one is part of the card.
Can we get NIST AI RMF certified?
No, and nobody can. Neither NIST nor CAISI runs a certification programme for the AI RMF - it is a voluntary framework, so there is no compliance state for anyone to certify against.
What does “aligned with the AI RMF” actually mean here?
That the engagement runs in the framework's four functions, that every finding names the category it answers, and that the mapping is published on this page for you to check rather than asserted in a deck.
Is the AI RMF mandatory for us?
Not by itself - it creates no legal obligation. What makes it feel mandatory is downstream: procurement questionnaires, boards, insurers and contract flow-downs increasingly ask for it by name.
What do you actually produce at the end?
An alignment profile across all four functions, an AI system register, a function-by-function map, a governance gap list, a scope memo, a ranked roadmap, and the hash-chained trail behind all of it.
How long does it take?
Typically two to four weeks from hand-over for a first profile, longer where the system count turns out to be bigger than expected - which it usually does. Scope is agreed before anything is charged.
Get started
Request an AI RMF assessment
Tell us about your AI systems and we come back with a scoping call within one business day.
What we need from you
Nothing you do not already have. Most of this is a folder someone can assemble in an afternoon, and we name every document first, in writing, before you commit.
- Which AI systems you build or use, and what each decides
- Any AI policy, however partial, and who signed it off
- Any documentation - model cards, contracts, DPIAs
- Whether you built the system, bought it, or modified one
- Your target readiness date, if you have one
What happens next
- You send the five items we need.
- You get a scoping call within one business day.
- Nothing is charged until you approve the scope.
Sources & standing
Where this page gets its facts
Where the claims on this page come from, and what they are worth - stated, not assumed.
What it is drawn from
- NIST AI 100-1 - the AI RMF 1.0
- NIST AI 600-1 - Generative AI Profile
- NIST AI 800-1 - 2nd public draft
- NIST AI 800-2 - initial public draft
- Published
- January 2023
- Revision due
- No date announced
What it means
- An alignment statement, not a certification — nobody certifies the AI RMF, and our mapping is self-attested. It determines nothing about your own systems.
- Where a mapping is arguable, our reports say so rather than the convenient thing.
Scope & limitation
- Do not rest a binding decision on it; engage qualified counsel.
- It covers AI RMF 1.0 alone - the EU AI Act and sectoral rules reach the same system.
- Use it as a starting point for a scoping conversation, not as your final word.
Something on this page out of date?
Tell usFrom Insights
Before you commission one
ISO/IEC 42001, SOC 2 and NIST AI RMF: Which One Your Buyer Is Actually Asking For
One certifies an organisation, one is an opinion about controls over a window, one is a method with nothing to issue. What each covers — and what none of them answers.
Read the guide
STANDARDS EXPLAINED
What an AI Governance Framework Actually Contains
Five working parts, not a policy document. What each one has to do, how to tell whether yours is real, and why a framework is not the same thing as compliance.
METHODOLOGY NOTES
How to Prepare for an AI Audit: The Readiness Checklist
Six things to have ready before the engagement starts. Assembling them takes a fortnight off the clock — and tends to find the first two findings before an auditor does.
METHODOLOGY NOTES
What Is an AI Audit? Scope, Standards, and What You Get
An independent review of what your AI actually does, measured against a named standard — not a certificate, and not a review of what the documentation says it does.
No payment until you approve the scope
Request an AI audit
Request received
Brijesh Patel, our founder, personally replies within one business day. Nothing is charged until you approve the scope.
- 1We reviewOur team confirms your scope, usually within one business day.
- 2You confirm & payLock it in securely to fast-track your slot.
- 3Your audit beginsUpload documents in your portal and we start.
We’ve emailed you a secure, password-free link to your client portal. Sign in to your portal →
For audit firms and practitioners: co-deliver AI audits on our methodology and audit trail.
Application received
Thank you — your auditor partner application is with our team. We review every application and will be in touch by email.