The PRA published SS1/23 in May 2023 and it took effect a year later. It reaches UK firms holding internal model approval, and sets out five principles: identification, governance, development and use, independent validation, and the mitigants that sit behind all four of them.
PRA SS1/23 · MODEL RISK · ANNUAL
If you hold an internal model permission, this is yours.
Live since 17 May 2024. Five principles for firms with internal model approval, and an annual self-assessment a senior manager owns.
Model inventory
Risk tiering
Independent validation
Post-model adjustments
Annual self-assessment
Our promise
“A principle is a heading. The model file 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 $25,000, and nothing is charged until you approve it.
- Each additional system
- $6,000
- Re-audit, same scope
- $16,000
- Renewal, every twelve months
- $21,000 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
SS1/23, in three chapters
Discovery is where these programmes lose time, every single time. The inventory is maintained rather than reconciled, the definition is drawn too narrowly to catch spreadsheets and vendor scores, and the tiering labels everything but changes nothing that follows from it.
Our review starts at the register and works outward from it. We reconcile the inventory against production, check that the tiering actually changes the treatment, test whether validation has standing to disagree, and leave the annual self-assessment in a state somebody can sign.
What counts as a model?
Wider than the things you call models.
A quantitative method turning input data into output — it expressly reaches deterministic methods and AI.
- Statistical, economic, financial and mathematical methods
- Deterministic methods — rules engines and spreadsheets
- AI and machine learning, inventoried like anything else
- Vendor and third-party models, on your own register
- IRB — credit risk, usually the largest population in the firm
- IMA — market risk, with the valuation models that feed it
- IMM — counterparty credit risk, on the same inventory
- No internal model permission — free to adopt it anyway
Whose duty is it
The model may be theirs. The model risk is yours.
The firm holding the permission
SS1/23 attaches to the firm with internal model approval, and inside that firm to a named senior manager. The inventory, the tiering, the validation, the post-model adjustments and the annual self-assessment are all yours, and none of them can be held by somebody else on your behalf.
The people who built the model
Carries no expectation of its own here - the statement is addressed to the firm. Many vendors document their models well and will supply what they have. The difficulty is not the quality of that work; it is that it stops at the edge of what they are willing to disclose, and that edge is theirs alone to set.
A bought model is your model risk
Vendor models belong on the inventory, get tiered like your own, and need enough documentation to validate and monitor. Where a supplier will not provide it, that is a finding against you rather than against them - so the hooks have to be written into the contract before anything is deployed.
What most firms assume
“It is a vendor model, so the documents are theirs.”
What the supervisor does
Asks you. A missing document is a finding on you.
It is the most common finding we write up.
- Who it is for
- Model risk management
- Independent validation
- Chief risk officers
- Credit & market risk
- Internal audit
- The accountable SMF
What weakness costs
02
Findings arrive through ongoing supervision and through the self-assessment you submit — remediated on the PRA’s timetable.
03
Model risk weakness can feed a Pillar 2A assessment or a scalar on the affected models. That is the one that shows up in the numbers.
04
In a persistent case: restrictions on model use, or a review of the internal model permission. For an IRB firm that is the outcome that bites.
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.
PRA SS1/23
Your scope check
The sequence
Four moments, and one of them repeats.
Nothing is submitted on a date, so the only thing that recurs is the one artefact somebody senior has to put a signature under.
-
In force
17 May 2024Published a year earlier and live since. It is the standard your supervisor has read you against for two years.
-
Self-assess
Every yearAgainst all five principles, with real gaps, owners and dates, reviewed by the accountable SMF and board.
-
Stress testing
23 April 2026The Low Impact Amendments bring stress test model risk practices inside exactly the same set of expectations.
-
Newly permitted
12 monthsA firm first granted internal model permission has a year from the grant to comply. The PRA confirmed it in 2026.
The trap
Every principle downstream of the register inherits its gaps in silence. Tiering, validation and monitoring are all blocked until the inventory can be trusted — so a programme that starts anywhere else spends its first quarter discovering that it has to start again.
Expectation & coverage
What the statement expects, what we ship
12 expectations, and the artefact that answers each one, in the PRA’s numbering.
- Firmwide model definition Principle 1 - deterministic and AI/ML included
- A definition your teams can apply, with worked calls on the borderline cases: spreadsheets, rules engines, vendor scores and AI-assisted steps.
- Model inventory Principle 1 - reconciled, not maintained
- A register reconciled to production, with owner, tier, purpose, status, dependencies and vendor provenance against every entry.
- Risk tiering methodology Principle 1 - materiality, complexity, uncertainty
- A tiering scheme applied across the estate, with the borderline calls documented and the treatment each tier actually receives written down.
- SMF accountability Principle 2 - a named senior manager
- The accountable senior manager identified, their responsibilities mapped, and the reporting line into the board evidenced rather than assumed.
- Model risk appetite Principle 2 - reported as a risk in its own right
- An articulated appetite with measures, so "within appetite" is a statement somebody can test rather than a sentence in a policy.
- Development standards Principle 3 - data, testing, documentation, change
- Data, testing, documentation and change-control standards, each with the evidence the stage has to leave behind.
- Validation methodology Principle 4 - independent, and scaled by tier
- A methodology scaled by tier, a findings register with owners and dates, and the independence argument written rather than asserted.
- Post-model adjustment governance Principle 5 - the quiet overlay problem
- A PMA register with justification, approval, expiry and a retirement plan for each adjustment currently in force.
- Monitoring and thresholds Principle 5 - including drift on AI and ML
- Metrics per model with thresholds, owners and an escalation route - and drift and data-quality checks that run unprompted.
- Vendor model governance Cross-cutting - an application of all five
- The documentation set to demand from a supplier, and the contractual hooks that make change notification and validation access enforceable.
- Annual self-assessment Cross-cutting - the artefact that proves it is alive
- The assessment against all five principles, with gaps, owners and dates, in a form the SMF can sign and the board can interrogate.
- Stress test model risk From 23 April 2026
- The stress-testing model population assessed against SS1/23 alongside the rest, following the PRA's April 2026 amendment.
The engagement
Your model estate, independently reviewed
From an IRB rating model to a scored spreadsheet.
-
Inventory
The model population reconciled to production, vendor and deterministic included.
-
Tier and test
Tiering applied, validation and monitoring checked against what each tier gets.
-
Sign off and self-assess
You see the draft first. Then the assessment the SMF can sign, gaps and dates.
Struck in your favour
Why firms choose iDharma for SS1/23
Genuinely independent
We build and resell no models, and we take no fee that is tied to what the review finds.
Written to the numbers
Every finding names the principle it answers, in the numbering your committee uses.
One review, three rules
SS1/23, SR 11-7 and OSFI E-23 are one estate and one piece of work, and not three of them.
We start at discovery
The inventory is where these programmes lose time, so it is where our scope always begins.
Four marks, struck on every review.
Deliverables
What you get
Concrete artefacts, each with a name and a format - you know what lands before you buy.
SS1/23 readiness review
The full review against all five principles, written in the PRA's own numbering so your committee reads it in the order it already thinks in: each one assessed with the evidence behind it, the cross-cutting vendor and self-assessment questions answered, and every gap carrying an owner and a date the SMF can sign under.
Model inventory
A register reconciled to production with owner, tier, purpose, status, dependencies and vendor provenance against every single entry.
Tiering methodology
The scheme, the factors behind it, the borderline calls documented, and the treatment each tier actually receives written down.
Validation methodology
Scope and depth scaled by tier, a findings register with owners and dates, and the independence argument written rather than asserted.
PMA register
Every post-model adjustment with its justification, approval, expiry and a plan to retire it rather than to keep it indefinitely.
Monitoring specification
Metrics per model with thresholds, severities and a configured action per breach - plus drift checks for the AI and ML population.
Annual self-assessment
The assessment against all five principles, comparable year on year, with gaps and dates the SMF can put a signature under.
Format & fee
Real numbers, upfront.
- Scope
- Set by the statement, not by us
- Input
- Your model estate and its records
- Re-assess
- Each self-assessment — $21,000 against your known baseline
The statement fixed the scope, so the fee is flat - nothing to meter, and nothing charged until you approve it.
Request this review
SS1/23 · Named engagement
$25,000 flat
- Inventory reconciled to production
- Tiering that changes the treatment
- Validation and PMA governance reviewed
- Annual self-assessment, ready to sign
Show your hand
Four things you have to be able to produce
These are a supervisor’s four questions. Each is either true of your framework on the day it is asked, or it is not.
The register,
reconciled
One inventory checked against production, with every entry owned - vendor models and the deterministic methods included, not kept on a separate list beside it.
The tier,
with teeth
A tier on every model, built from materiality, complexity and uncertainty, and visibly changing how much validation and monitoring each one gets.
The stop,
on record
Validation in date on Tier 1, findings tracked to closure, and at least one occasion on which it stopped something going out of the door. That last part is the real test.
The paper,
honest
A self-assessment against all five principles with real gaps, owners and dates. One that comes back with no findings is read as one that was never performed at all.
Four cards, and the date on each one is part of the card.
FAQ
Plain answers
Reach, what counts as a model, tiers, weakness. Answered straight.
Request this reviewWho does SS1/23 apply to?
UK-incorporated banks, building societies and PRA-designated investment firms that hold internal model approval - IRB, IMA or IMM. If you do not hold one of those permissions, the supervisory statement does not bind you.
What counts as a model?
The definition is deliberately wide, and it expressly reaches deterministic methods and AI and machine learning - so a rules engine or a scored spreadsheet can be squarely inside it.
Does SS1/23 require three tiers?
No. It requires classification proportionate to materiality, complexity and uncertainty, and expects it to drive the intensity of the controls. Three tiers is what most firms land on, not a requirement.
What happens if our framework is weak?
The PRA does not fine for this. The consequences escalate: supervisory findings, then a capital consequence through Pillar 2A or a model scalar, and in a persistent case restrictions on model use or a review of the permission.
How does SS1/23 compare with SR 11-7 and OSFI E-23?
They are the same object for three supervisors. SS1/23 is distinctive for its explicit SMF accountability, its treatment of post-model adjustments, and the annual self-assessment. A firm running any one is most of the way to the others.
Get started
Request an SS1/23 readiness review
Tell us the shape of the model estate and we come back within one business day.
What we need from you
Nothing you do not already have. Most of this comes out of your model governance records in an afternoon, and we tell you exactly which extracts before you commit.
- Which internal model permissions you hold
- Roughly how large the model population is
- Your model inventory, if one already exists
- How much of it came from a vendor, and what they sent
- Whether AI or ML models sit in the population
What happens next
- We agree the scope with you first.
- Four to eight weeks, longer for a large inventory.
- 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
- SS1/23 and PS6/23, Bank of England
- Low Impact Amendments, April 2026
- In force
- 17 May 2024
- Amended
- 23 April 2026
What it means
- General information about what the statement expects — not regulatory advice, and no professional relationship.
- Where a scope question is genuinely arguable, our reports say so rather than pick the convenient answer.
Scope & limitation
- Drawn from the Bank of England’s own publications rather than from summaries, but not line-checked against SS1/23 or the April 2026 amendment. No penalty figure appears anywhere on this page, because the PRA does not fine for this.
- On an engagement we work from the statement itself. Use this 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
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.
Read the notes
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.
STANDARDS EXPLAINED
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.
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.
