NEW APRA letter to industry · 30 April 2026

The independent AI audit. Issued,
not advised.

kn0w is the independent AI audit for FinTech and HealthTech. One fixed-scope engagement, delivered entirely by software, issued as a signed Opinion the people you answer to can rely on — and reissued each year under the same methodology.

app.kn0w.co / opinions / NNNNNN
Opinion · kn0w / NNNNNN / YYYY-MM-DD / vX.Y
Issued
Member · FinTech · AU
AI Accountability Posture — Annual
Quartile
Q2
D1 · Workflow Automation
62
D2 · AI Tool Deployment
54
D3 · AI Literacy
48
D4 · Governance & Oversight
71
D5 · AI Investment Time
58
D6 · Outcome Tracking
44
kn0w · Hallmarkkn0w / NNNNNN / YYYY-MM-DD / vX.Y
Frameworks the Opinion maps to
APRA
ASIC
FCA
TGA
MHRA
CPS 230
SS 4/26
EU AI Act
02 — Why this matters now

The regulator isn't asking whether your AI is good. They're asking what evidence you have that it was governed.

APRALetter to Industry on Artificial Intelligence
"APRA observed internal audit and risk functions are challenged. Many lack the specialist skills and tools required to engage in AI assessment or audit. This is particularly true where agentic behaviour, automated decision making or AI assisted code generation were involved."
apra.gov.au · 30 April 2026

APRA, ASIC, the FCA, the TGA, the MHRA. Eleven AU and UK frameworks now hold named senior executives personally accountable for AI governance.

The test isn't whether your AI works. The test is whether you can show, in writing, in advance, on a document a regulator recognises, that you took reasonable steps.

Most companies submit a slide deck made for a board meeting, an AI policy pulled together over a quarter, or a memo a law firm wrote after the fact. None of those documents was built to be submitted, benchmarked, or signed by an independent party with no services revenue on the line.

No existing document is built to be submitted. kn0w is.

11
AU & UK frameworks naming personal accountability
5
Regulators in scope — APRA, ASIC, FCA, TGA, MHRA
4wk
From contract to signed Opinion in your hand
k=5
Anonymisation threshold across the peer cohort
03 — How kn0w is different

The fourth option.

Unaccountable AI is the condition where AI is deployed across a regulated company with no record of who is on the hook when it fails. It's the condition every framework above is written to close. You have three ways to close it. None is independent of what they're being asked to measure.

Advisory
The Consultant
Conflicted
Sells advisory hours. Recommends what they're billing you to implement.
Software
The Vendor
Conflicted
Sells software subscriptions. Grades the homework they sold you.
Billable
The Lawyer
Conflicted
Sells billable hours. Writes the memo after the fact, in defensive language.
ConsultantVendorLawyerkn0w
Independent of what it measures
Fixed-scope, fixed price
Benchmarked against a named peer cohort
Issued as a signed Opinion, re-issued annually

The Opinion is benchmarked against a cohort of named peers — FinTech or HealthTech, 50 to 200 staff, in AU or UK under APRA, ASIC, FCA, TGA, or MHRA oversight. Not an industry average. Not a synthetic comparison set. A cohort, anonymised at k=5 from the first audit, that no other firm is structurally permitted to hold.

04 — What you walk away with

A signed document, in your hand, in four weeks.

The audit runs over four weeks. Seven voice sessions — your CEO and the heads of your six core functions — delivered by software. At the end, kn0w issues your Opinion: a signed document carrying a reading of your AI accountability posture across six dimensions.

What the Opinion measures

Six weighted dimensions. One scoring layer. Each mapped to the named regulatory instruments your company answers to.

D · 01
Workflow Automation Rate
How much of the work has been moved to software — and where AI sits inside it.
D · 02
AI Tool Deployment
Which AI systems are in production, who runs them, what they're authorised to do.
D · 03
AI Literacy
Whether the people accountable for AI can describe what they're accountable for.
D · 04
Governance & Oversight
The chain of human review, escalation, and sign-off behind every AI decision.
D · 05
AI Investment Time
Where senior time is going — implementation, oversight, vendor management, training.
D · 06
Outcome Tracking
What you measure after deployment, and what triggers a rollback.
§
Not a rating. A reading. Issued on a date, under a methodology version, against a peer cohort — not a continuous score, not amended after the fact, not a guarantee.The institutional signature · kn0w / NNNNNN / YYYY-MM-DD / vX.Y
Inside the Opinion

Six components. One issued document — the founding act of a standing cycle, reissued each year under the same methodology.

01
Six-dimension reading
A reading of your AI accountability posture across six dimensions. Composite and per-dimension. Reported as a Quartile band with a cohort percentile.
02
CEO Visibility Gap
The signed delta between what the CEO sees and what the function heads report, dimension by dimension, headcount-weighted.
03
Regulatory mapping
Your readings routed to the regulatory frameworks your company answers to. Eleven AU and UK frameworks in primary scope. Each obligation stamped with its status.
04
Quartile placement
Your placement against the peer cohort — FinTech or HealthTech, 50–200 staff, AU or UK. Anonymised at k=5 from the first audit.
05
90-day required actions
The priority actions for the next twelve months — named, sequenced, owned. Findings, not recommendations.
06
Verification status
Every regulatory claim traced to a primary source, corroborated by secondary where the primary was inaccessible, or flagged as interpretation where the primary source is silent.

You can submit it to your board. You can submit it to your regulator. You can put it in a diligence data room when your next round asks about AI governance.

05 — How payment works

Fixed price. One contract.

No retainer. No success fees. No scope creep. One price to commission your Opinion.

Lifecycle mechanics, for active members

Annual standingeach year on the anniversary
A$24,000
Material-change re-Audittriggered by material change events
A$17,500
06 — Who this is for

kn0w is built for a specific kind of company.

01
Sector
FinTech or HealthTech regulated under APRA, ASIC, FCA, TGA, or MHRA.
02
Scale
50 to 200 staff. Past the founding team, before institutional scale.
03
Geography
Australia or the United Kingdom. Or both — the benchmark holds across jurisdictions.
04
Posture
A live AI governance question at senior level. If you're not facing one, you're not ready.

The audit is built to be submitted.
Not explained.

Your board asks how AI is governed. Your regulator asks for evidence of reasonable steps. Your next round asks about AI risk in diligence. The answer to all three is the same document.