As at August 2026. General commentary on publicly available regulatory and professional frameworks, not legal advice. The UK accountability regime is under active reform; verify current status against FCA, PRA and professional-body sources.
Model risk accountability roles answer the question every model failure eventually forces: who, by name, was responsible? The UK and the US answer it through strikingly different architectures - the UK through a statutory individual-accountability regime layered over professional duties, the US through institutional supervision plus professional standards enforced by the profession itself. Neither architecture transfers accountability to a vendor, a committee or an algorithm, but they locate it differently, and groups operating in both jurisdictions routinely discover that their accountability maps do not line up.
The UK architecture: named individuals, statutory teeth
The Senior Managers and Certification Regime - in force for banks since March 2016 and for dual-regulated insurers since December 2018 - is the spine. Senior Management Functions are held by named, approved individuals; each has a statement of responsibilities; and the duty of responsibility means a senior manager can face regulatory action if a failing occurred in their area and they did not take reasonable steps. Model risk sits inside this structure explicitly: UK supervisory expectations for banks identify an accountable senior manager for the model risk framework, and model-driven failures - mispriced risk, misstated capital, discriminatory outcomes - land in someone's statement of responsibilities.
The regime is being reshaped: reforms announced in April 2026 aim to streamline it, including the proposed removal of the Certification Regime in favour of firm-led fitness assessment, with further consultation to follow. Parliamentary scrutiny has run the other way too - a Treasury Committee report in early 2026 pressed on whether SM&CR gives adequate assurance when AI systems cause consumer harm, foreshadowing pressure to make algorithmic accountability more explicit rather than less. The direction of reform is simplification; the direction of expectation, for models, is not.
Beneath the statutory layer sits the professional one: actuaries hold practising certificates for reserved roles, are bound by the Actuaries' Code, and work within FRC technical actuarial standards - so a UK actuary signing model-dependent work answers to three masters at once: regulator, standards-setter and profession.
The US architecture: institutional supervision, professional self-regulation
US banking regulation supervises institutions: model risk expectations under the revised interagency guidance are enforced through examination, MRAs and enforcement actions against firms, with individual liability generally reserved for misconduct rather than framework failings. There is no general US analogue of the SM&CR's pre-approved, named-function architecture.
The individual layer in US financial services is professional, and strongest in actuarial work. The Appointed Actuary signs statutory Statements of Actuarial Opinion on reserves - a named, personal attestation. The US Qualification Standards of the American Academy of Actuaries govern who may issue statements of actuarial opinion; Actuarial Standards of Practice issued by the Actuarial Standards Board govern how the work is done - including reliance on models and on others' work; and the Actuarial Board for Counseling and Discipline investigates and recommends discipline. The result: for reserving and solvency opinions, US accountability is as personal as anything in the UK - but it attaches through the profession rather than the regulator, and it covers actuaries, not the full population of model owners.
Where the maps fail to overlap
The quant gap. A machine-learning pricing model built by non-actuaries sits, in the UK, under a senior manager's statement of responsibilities regardless of who built it. In the US, if it is outside actuarial opinion scope and below enforcement thresholds, no individual regime reaches it at all - only institutional supervision.
The reliance chain. Both systems require the accountable person to have done something real: reasonable steps in the UK; under US actuarial standards, disclosed and appropriate reliance. Neither permits "the model said so" - a point I have developed for AI specifically in my article on LLM actuarial reasoning: reliance you cannot independently defend is substitution, not support.
Group blind spots. A US-parented group may run UK entities with an SM&CR map that is formally complete but thin on model risk substance; a UK-parented group may assume its US operations carry named accountability that in fact exists only for actuarial opinions. The gap analysis is worth an afternoon and occasionally prevents a career-defining surprise.
The practical discipline
Whatever the jurisdiction, the durable practice is the same: a model-to-owner map covering the whole inventory, with one named individual per material model; owners equipped to discharge the role - validation evidence they can read, monitoring they actually receive, authority to restrict use; and documented reliance: who relied on whom, for what, on what basis. Regimes differ in where they put the name. Good firms make sure there is always a name.
Key Takeaways
- The UK locates model accountability in named senior managers under SM&CR, backed by professional duties; the US supervises institutions, with personal accountability arising mainly through actuarial professional standards.
- SM&CR reform (April 2026 announcements, including proposed removal of the Certification Regime) streamlines the machinery, while parliamentary scrutiny presses for clearer accountability when AI systems cause harm.
- US Appointed Actuaries, Qualification Standards, ASOPs and the ABCD create personal accountability for actuarial opinions - but non-actuarial model owners face no equivalent individual regime.
- Dual-jurisdiction groups should map models to named owners across both regimes; the common failure is formal completeness with substantive gaps.
- As at August 2026; both regimes are moving - verify before reliance. Not legal advice.
Frequently Asked Questions
Who is accountable for model risk under SM&CR? The senior manager in whose statement of responsibilities the relevant activity sits - for banks, supervisory expectations identify an accountable senior manager for the model risk management framework specifically. The duty of responsibility exposes that individual to regulatory action if reasonable steps were not taken.
Who is responsible for models in US insurance? Institutionally, the insurer under state regulation and, for banking organisations, the federal supervisors. Personally, the Appointed Actuary for statutory opinions, governed by the Academy's Qualification Standards and the ASB's standards of practice, with discipline through the ABCD - a professional rather than regulatory accountability chain.
Can model accountability be delegated to a vendor? No, in either jurisdiction. UK senior managers retain responsibility for outsourced and vendor-supplied models under their remit, and US actuarial standards require disclosed, justified reliance. Contracts can allocate cost; they cannot relocate regulatory or professional accountability.