SECTOR INTELLIGENCE BRIEFING, BASED ON DOCUMENTED PUBLIC EVIDENCE, RESEARCH DOCUMENT
01The Documented Picture
◈ This briefing draws exclusively on documented, publicly available evidence from UK regulators, parliamentary committees, and authoritative surveys. All claims are citable. The predictive findings in Section 04 are inferences from this evidence base, not assertions of findings BreakPoint has made in this sector.
95%
of insurance firms already using AI
BoE/FCA Survey, Nov 2024
46%
have only partial understanding of their AI
BoE/FCA Survey, Nov 2024
55%
of AI use cases include automated decision-making
BoE/FCA Survey, Nov 2024
84%
have a named accountable person for AI frameworks
BoE/FCA Survey, Nov 2024
Insurance is the UK financial services sector with the highest rate of AI adoption, 95% of firms surveyed by the Bank of England and FCA in November 2024 are already using AI. The sector uses AI across claims triage, underwriting risk scoring, fraud detection, pricing, and customer communications. The Treasury Select Committee, in its January 2026 report on AI in financial services, specifically cited "processing insurance claims" as one of the core services now being delivered by AI, not merely supported by it.
The governance picture is significantly less advanced than the deployment picture. High materiality AI use cases are disproportionately concentrated in general insurance (BoE/FCA Survey 2024). The same survey found that 46% of firms have only a "partial understanding" of the AI they use, principally because a third of all AI deployments are now third-party implementations whose internal logic firms cannot fully examine. This creates a documented gap between governance claims and governance reality that is characteristic of the sector.
02The Governance Gap, What the Evidence Shows
The Treasury Select Committee's January 2026 report found that the FCA, Bank of England, and HM Treasury are "not doing enough to manage the risks presented by AI." Insurance, alongside international banking, was specifically identified as having the highest AI adoption. The Committee's concern: AI is delivering core services including claims processing in a regulatory environment that has not yet determined who is accountable when those AI-shaped outcomes harm consumers. The report explicitly called for FCA guidance on how Consumer Duty applies to AI and on SMCR accountability for AI harm, to be published by the end of 2026. At the time of writing, that guidance does not yet exist.
The most cited governance metric in the BoE/FCA 2024 survey is that 84% of firms have an "accountable person or persons with responsibility for the AI framework." This figure is widely reported as a positive governance indicator. It is not. Having a named accountable person at framework level is the minimum governance architecture. What it does not evidence, and what neither the survey nor any regulatory guidance currently measures, is whether that person exercises authority over AI-shaped decisions at decision level. The gap between "named for the framework" and "evidenced in the decision" is precisely the gap BreakPoint is designed to determine.
Third-party AI implementations account for 33% of all AI use cases in the financial sector, up from 17% in 2022. In insurance, where operational AI is heavily embedded in claims and underwriting workflows, vendor-supplied models that firms cannot fully interrogate are now a material governance risk. The same survey found that partial understanding is "primarily due to the reliance on third-party models." Governance frameworks that assert human oversight of AI models the firm cannot fully understand are asserting oversight they may not be able to exercise in practice.
FCA Consumer Duty requires firms to demonstrate that their products and services deliver good outcomes for customers. The FCA has confirmed that this obligation applies where AI influences customer-facing decisions, including claims handling, underwriting acceptance, and customer communications. The duty is outcomes-based: firms cannot satisfy it by pointing to a policy framework or a governance document. They must be able to evidence that the outcomes AI-influenced decisions produced are fair and appropriate. The FCA's 2024 AI Update confirmed that Consumer Duty is the primary lens for assessing customer-facing AI in insurance. No guidance on how to evidence this in practice currently exists.
PRA Supervisory Statement SS1/23, effective from May 2024, establishes model risk management as a risk category on par with credit and market risk. It applies to all models used to inform business decisions, including third-party and vendor models. PRA-regulated insurers with AI-driven underwriting scoring models are expected to have independent model validation, senior management capable of challenging model assumptions, and an ongoing validation schedule. The statement explicitly notes that AI and machine learning models introduce additional complexity around explainability, data provenance, fairness, and accountability that standard accuracy metrics do not capture.
03The Regulatory Framework, What Applies and What It Requires
FCA Consumer Duty (PS22/9)
Firms must demonstrate that AI-influenced decisions deliver good customer outcomes, not merely that governance documents say they do.
Claims triage, underwriting acceptance/denial, pricing, customer communications. Any AI-shaped outcome that affects a consumer.
FCA SM&CR
Accountability for AI-influenced decisions sits within existing Senior Manager Functions, no dedicated AI SMF. The Treasury Select Committee has called for clarity on what this means in practice.
Chief Operations Officer, Chief Risk Officer, or function head with AI-active processes in scope. Accountability certification without decision-level exercise is the documented gap.
PRA SS1/23 (PRA-regulated insurers)
Model risk on par with credit risk. Independent validation required. Senior management must be able to challenge AI model assumptions. Third-party models included in scope.
Underwriting scoring models, pricing AI, reserving models. Validation must go beyond accuracy to include transparency, bias, and fairness. Applies to vendor models the firm relies on.
ICO, Algorithmic Decision-Making
Where AI makes or significantly influences decisions about individuals, firms must be able to explain those decisions and provide a meaningful human review route.
Claims decisions, underwriting acceptance, fraud flags. Human review must be meaningful, not a sign-off within the AI's pre-determined decision space.
FCA PROD Rules, Product Governance
Products must be designed and monitored to deliver fair value. Where AI shapes product distribution or claims outcomes, product governance must cover the AI's role.
Customer segmentation AI, pricing optimisation, targeted product offers. Scope creep into product governance territory is a documented risk pattern.
04What BreakPoint Would Predict, Specific Gaps a BreakPoint Assessment Would Expect to Surface
Epistemic note: The following are indicative and predictive inferences from the documented evidence base above - not assertions of findings BreakPoint has made in this sector. Each prediction identifies a specific gap type that the evidence strongly suggests is present. BreakPoint's role is to determine, for a specific organisation, whether the gap exists and its severity. Note that BreakPoint is not limited to these factors however, and will often surface unexpected risks or risks that were not fully articulated.
Where AI categorises incoming claims, determining pathway, priority, and assessment framework, the evidence strongly suggests that no named individual holds documented authority to challenge or override the categorisation at the point it is made. The claims function head is typically named as the authority holder at framework level; the operational question BreakPoint asks is whether that authority is exercised before triage consequence, not reviewed after it.
Evidence basis: BoE/FCA Survey 2024 (84% framework accountability, 0% documented decision-level exercise); Treasury Select Committee 2026 (claims AI delivering core services without adequate accountability clarity); FCA Consumer Duty (outcomes must be evidenced, not asserted).
AI underwriting scoring tools are consistently described as "advisory" or "decision-support" in governance frameworks across the sector. The BoE/FCA survey evidence, that 55% of AI use cases include automated decision-making and that firms have only partial understanding of their AI, is consistent with advisory claims that are not substantiated by override rate measurement. A tool described as advisory but not measured for override rate is a tool whose advisory status is asserted but unevidenced. BreakPoint would measure the override rate.
Evidence basis: BoE/FCA Survey 2024 (55% automated decision-making; 46% partial understanding); PRA SS1/23 (model validation must go beyond accuracy metrics); FCA Consumer Duty (outcome evidence required, not assertion).
The documented gap between governance framework content and operational reality is a sector-wide pattern evidenced by the BoE/FCA survey (46% partial understanding; third-party models as a primary source of opacity). Governance documents that describe human oversight of AI processes the firm only partially understands describe an intention, not an operation. BreakPoint would test the gap between what the framework says and what the decision record shows.
Evidence basis: BoE/FCA Survey 2024 (46% partial understanding; 33% third-party implementations); Treasury Select Committee 2026 ("wait and see" approach failing to keep pace with AI deployment).
Where AI influences a claims or underwriting decision that subsequently generates a complaint or regulatory challenge, firms are unlikely to be able to identify the AI's role in the causal chain from existing complaints and remediation records. This is both a governance blindspot and a Consumer Duty exposure: firms cannot evidence good outcomes by root cause if their records do not capture which outcomes had AI in their causal chain.
Evidence basis: FCA Consumer Duty (outcome evidence required); BoE/FCA Survey 2024 (monitoring metrics focus on model accuracy, not outcome attribution by AI causation); Treasury Select Committee 2026 (accountability for harm caused through AI).
AI-generated customer communications are a documented growth area in insurance. The approval governance for communications AI typically covers the scope at point of deployment. Scope extension into communications that touch policy terms, claims decisions, or regulatory disclosures, without corresponding governance update, is a pattern consistent with the documented sector-wide gap between AI deployment pace and governance adaptation pace.
Evidence basis: BoE/FCA Survey 2024 (customer support identified as the biggest expected growth area for AI in next three years); FCA Consumer Duty (communications must support informed decisions); ICO algorithmic decision-making rules.
05Source References
BoE/FCA 2024
Bank of England / Financial Conduct Authority,
Artificial Intelligence in UK Financial Services 2024, November 2024. Third annual survey. 118 respondents across six sectors.
bankofengland.co.uk →Treasury Committee 2026
House of Commons Treasury Select Committee,
Artificial Intelligence in Financial Services, January 2026. Finding: current regulatory approach risks serious harm to consumers. Recommendations include FCA guidance on Consumer Duty/AI and SMCR accountability for AI harm.
parliament.uk →PRA SS1/23
Prudential Regulation Authority,
Model Risk Management Principles for Banks, Supervisory Statement SS1/23, May 2023. Effective May 2024. Model risk on par with credit risk; independent validation; senior management challenge capability required. Applies to third-party models.
bankofengland.co.uk →FCA Consumer Duty
Financial Conduct Authority, Consumer Duty Final Rules and Guidance, PS22/9, July 2022. Effective July 2023. Outcomes-based requirement. FCA confirmed application to AI-influenced decisions in 2024 AI Update.
FCA AI Update 2024
Financial Conduct Authority,
AI Update, 2024. Confirms no AI-specific rules; Consumer Duty and SM&CR are primary frameworks.
fca.org.uk →FCA Dear CEO 2024
Financial Conduct Authority,
Insurance Product Governance, Dear CEO Letter, February 2024. Supervisory priorities for the insurance sector including product governance and consumer outcomes.
insurancecompliance.co.uk →
Luminary Diagnostics · BreakPoint: AI Control Gap Assessment
Insurance has 95% AI adoption.
The evidence of governance keeping pace is not there.
BreakPoint determines - for a specific firm, from operational evidence - where the gap between documented governance and actual human authority exists, and what it means for regulatory exposure.
Book a BreakPoint conversation →