The FCA Mills Review has thrust insurance pricing algorithms into the regulator’s AI risk radar, warning that hyper-personalised pricing can drift into opaque pricing and personalised manipulation that erodes fair value for policyholders. Published in early July 2026 and led by FCA executive director Sheldon Mills, the review is billed as the first work of its kind initiated by a regulator globally. For insurers, the message is blunt: the FCA is not waiting for bespoke AI rules — it expects Consumer Duty price-and-value testing to catch algorithmic pricing risk now.
Sheldon Mills’s First-of-Its-Kind AI Review Lands After 140 Submissions
Commissioned by the FCA Board and published as a standalone corporate document rather than a consultation, the Mills Review sets out how artificial intelligence could reshape UK retail financial services — including insurance — by 2030, the FCA said in the press release accompanying the review. The review draws on 140 written submissions plus meetings, panels and roundtables spanning financial services, technology, academia, consumer groups and regulators in the UK and abroad, according to the published review. Mills himself frames the pace of change as the reason for urgency: more than 20 frontier AI models, plus hundreds of smaller variants, have been released since he started the project in late 2025.
Where the Review Draws the Line: “Opaque Pricing” and Fair Value
The insurance-specific warning sits inside the review’s discussion of hyper-personalisation. AI-driven personalisation, the review argues, could help insurers match products to customer needs, but the same technology could enable bias, opaque pricing and personalised manipulation. That risk is mapped directly onto the Consumer Duty’s price-and-value outcome: the review states that AI-enabled pricing may make it harder to distinguish benign personalisation from extraction where consumers pay more without receiving additional value — precisely the test insurers must already pass under the Consumer Duty’s requirement to provide products and services that meet customer needs and deliver fair value.
The review also flags a fairness risk beyond price: as AI increasingly shapes product design and delivery, it may become harder to ensure outcomes stay fair across different consumer groups, particularly where models embed historic patterns of bias. For insurance underwriting and rating models — which already draw scrutiny from the House of Lords’ probe into UK insurance pricing — that is a direct line from AI governance to pricing fairness.
The Protection and Advice Gaps AI Is Meant to Close
The review is not solely a warning shot. It frames AI’s upside in insurance around gaps the industry has struggled to close through conventional means. Only 30% of UK consumers hold life or income protection insurance, a protection gap the review says AI-enabled tools could help narrow, alongside an advice gap in which just 9% of consumers currently use traditional regulated financial advice. Financial exclusion features too: around 900,000 people in the UK remain unbanked, per the review’s own figures.
FCA-commissioned research cited in the review found a fifth of people — equivalent to 11 million UK adults — are likely to use AI that can act autonomously within pre-set financial goals. The review’s illustrative example of “agentic finance” includes an AI agent that could identify better insurance products, manage claims or complaints, and switch providers on a consumer’s behalf — the same delegation dynamic that, per the review, could in theory change pricing, products or services in markets such as savings and insurance. Yet trust in unregulated tools is already running ahead of consumer protections: around 26% of consumers say they trust general-purpose AI tools such as ChatGPT, Claude or Gemini for financial advice, often unaware that formal complaint routes do not apply.
No New AI Rulebook — Consumer Duty and SMR Carry the Weight
For compliance teams, the review’s most consequential line may be what it does not propose. The FCA is not creating an insurance- or AI-specific rulebook. FCA Chair Ashley Alder said in the press release accompanying the review that the FCA’s principles-based, outcomes-focused approach to AI — relying on the Consumer Duty and Senior Managers Regime — has been critical to keeping pace with a rapidly changing environment. In practice, that means the pricing and fairness risks the review identifies are supervised through existing obligations: Consumer Duty price-and-value testing, and individual accountability for AI outcomes under the Senior Managers Regime.
The FCA is already testing that approach in practice. Its second cohort of AI Live Testing, announced on 21 April 2026, comprises eight firms exploring risk management and live monitoring for responsible AI deployment — a sandbox-style mechanism the regulator can extend to insurers experimenting with pricing algorithms without writing new binding rules.
How the UK’s Light-Touch Stance Contrasts With Brussels’ AI Mandate
The governance-now approach puts the FCA on a different track from the EU. EIOPA has moved toward a prescriptive AI mandate for insurers, with a two-step compliance deadline and substantial penalties for insurers that fail to meet its AI governance requirements. The FCA, by contrast, is betting that Consumer Duty and SMR are flexible enough to catch AI pricing harms without a comparable rulebook — even as its own survey work shows consumer scepticism toward algorithmic pricing echoes findings elsewhere in Europe, where only a minority of Europeans say they would trust AI-set insurance prices. Whether principles-based supervision can hold that line as agentic AI adoption accelerates is likely to be tested well before the review’s 2030 horizon.
