AI in Insurance: How Algorithms Are Calculating Your Risk in the UK

Insurance companies have always been in the data business. For 300 years, they collected information about risk — your age, your health, your driving record — to calculate what you should pay. Artificial intelligence is now fundamentally changing that process. UK insurers are using machine learning to predict risk more accurately than ever before, and the implications for your premiums, your privacy, and your legal rights are profound. This article explains what is happening, who is doing it, and what the FCA rules actually say about your protection.

How Insurance Has Always Used Data — and Why AI Changes Everything

Traditional insurance pricing relies on actuarial tables — statistical models built from decades of historical claims data. An actuary might calculate that drivers aged 18 to 25 in London present a higher accident risk than older drivers in rural areas, and price policies accordingly. These models are effective but blunt. Every young London driver pays a similar premium, regardless of how carefully they personally drive.

AI changes this at a fundamental level. Machine learning models can process thousands of variables simultaneously — far more than any human actuary can handle. Instead of broad demographic categories, AI-powered insurers build personalised risk profiles using real-time behavioural data. The pitch is more accurate pricing for more customers. The question is which data insurers are permitted to collect, and what happens when the algorithm produces a discriminatory outcome.

The Association of British Insurers estimates that 90% of UK motor insurers now use some form of machine learning in their pricing models, up from under 30% in 2019. Home insurers, life insurers, and commercial lines are following rapidly. This is not a future trend. It is the current state of the UK insurance market.

Telematics: The Car Insurance Revolution Already Happening

Telematics — colloquially known as black box car insurance — is the most visible example of AI-driven insurance in the UK. Companies including Aviva, Direct Line, Admiral, and specialist providers like Marmalade and ingenie have sold telematics policies since the early 2010s. A small device plugged into your car’s OBD-II diagnostic port, or a smartphone app, records your speed, acceleration, cornering, braking, and time of driving. This data feeds a machine learning model that continuously updates your risk score.

By 2025, over 1.5 million UK drivers had active telematics policies, according to the British Insurance Brokers’ Association. These policies are particularly popular with young drivers, where standard premiums frequently exceed £3,000 per year in major cities. Drivers who demonstrate consistently safe behaviour — avoiding harsh braking, observing speed limits, limiting late-night journeys — can save up to 40% compared to standard market rates. The AI rewards proven behaviour rather than statistical demographic assumptions.

The surveillance dimension cannot be ignored. Your insurer’s system knows where you drive, at what times, how fast you corner, and how sharply you brake. This data is commercially valuable well beyond insurance pricing. Some telematics providers have faced scrutiny over data-sharing arrangements with third parties including vehicle manufacturers and mapping services. Under UK GDPR, insurers must obtain explicit consent for any secondary use of telematics data and cannot share it beyond the stated purpose without fresh permission.

AI in Home and Contents Insurance

Home insurance underwriting has been transformed by satellite imagery and environmental data. Machine learning models now analyse aerial and satellite images to assess individual property roof condition, subsidence indicators, drainage infrastructure, and proximity to flood risk zones at street level. Companies including JBA Risk Management and Ambiental supply this granular risk data to insurers. Properties within 200 metres of a historically active flood plain may face significantly higher premiums even if the specific building has never flooded.

Smart home devices represent the next major frontier. Several UK insurers already offer premium discounts of 10% to 15% to households fitted with smart security systems, smoke detectors linked to monitoring services, and water leak sensors connected to the insurer’s partner app. The commercial logic is straightforward — real-time risk monitoring enables early intervention that prevents claims. If your leak sensor triggers, the insurer’s connected system alerts you before a small drip becomes a claim worth tens of thousands of pounds.

Critics note this creates a two-tier insurance market. Households with older properties, renters in accommodation they cannot modify, or those who cannot afford smart devices may face higher premiums by default — not because of their actual risk behaviour but because of their inability to participate in data-sharing schemes. The FCA’s Consumer Duty framework, introduced in July 2023, requires insurers to demonstrate fair value for all customers, including those who decline to share additional data.

Health and Life Insurance: What Your Data Could Reveal

Health and life insurance present the most sensitive AI applications in the UK insurance sector. The widespread adoption of wearable devices — Apple Watch, Fitbit, Garmin fitness trackers — generates detailed physiological data that insurers find commercially interesting. In the US and Australia, major life insurers including John Hancock and AIA have offered substantial premium discounts in exchange for sharing fitness tracker data. UK insurers are watching closely but moving more cautiously.

In 2025, Zurich UK launched a pilot programme offering life insurance customers premium reductions in exchange for sharing anonymised health metrics from wearable devices. The programme attracted 40,000 participants in its first year. Zurich argues the scheme reduces reliance on invasive medical examinations — a genuine benefit for many customers who struggle with traditional underwriting processes. Critics argue it creates commercial pressure on policyholders to share deeply personal biometric data as an implicit condition of affordable life cover.

The Genetics and Insurance Committee (GAIC) regulates the use of genetic test results in UK insurance underwriting. Currently, insurers cannot require applicants to undergo genetic testing or use known genetic results in underwriting decisions for policies under £500,000. AI raises novel questions about what constitutes “genetic” information in practice. If an AI model infers significant health risk from purchasing patterns, postcode, social media activity, or wearable heart rate variability data, is that functionally equivalent to using a genetic marker? This question has not yet been definitively resolved in UK case law.

The FCA’s Role in Regulating Insurance AI

The Financial Conduct Authority oversees the conduct of all FCA-authorised UK insurers and has taken an increasingly active interest in algorithmic underwriting. In January 2022, the FCA banned “price walking” — the practice of charging long-standing customers more than new customers for equivalent cover. This decision was driven in part by evidence that AI pricing models were systematically exploiting customer inertia, with loyal customers effectively subsidising acquisition discounts for new policyholders.

The Consumer Duty, which came into force in July 2023, goes considerably further. It requires all financial firms to act in good faith towards retail customers, avoid causing foreseeable harm, and actively support customers in achieving their financial objectives. Applied to AI-driven insurance underwriting, this means firms must be able to explain the factors that have produced a particular premium, demonstrate that pricing is fair across different customer groups, and take action when their models produce outcomes inconsistent with good value.

In February 2025, the FCA published specific guidance addressing AI in insurance underwriting and pricing. The guidance confirmed that existing regulatory requirements apply equally to AI-generated decisions as to human-made ones. It reminded firms of their obligations under the Equality Act 2010 — an AI model that produces discriminatory pricing outcomes based on race, sex, disability, religion, or other protected characteristics is unlawful, regardless of whether the discrimination was intentional. Firms are required to test their pricing models for bias against protected characteristics before deployment.

What AI Means for Insurance Premiums

For many customers, AI-driven insurance pricing produces genuine benefits. Drivers who would previously have been penalised simply for their postcode or age demographic can now demonstrate safe driving behaviour and access meaningfully lower premiums. Homeowners in verifiably low-risk areas may see more accurate pricing that reflects their actual exposure rather than a blunt regional average. The market is becoming more granular, which tends to reduce cross-subsidisation between high and low-risk customers.

However, more accurate risk pricing carries a significant distributional risk. As insurers become better at identifying genuinely high-risk customers, those customers may find it increasingly difficult to obtain affordable cover. A homeowner whose property sits in a verifiable flood zone, or a driver with a history of multiple fault accidents, may face premiums that effectively exclude them from the market. The government’s Flood Re scheme — a reinsurance pool that caps flood-risk home insurance premiums for eligible properties — exists precisely because the private market alone would leave many UK homeowners uninsurable at reasonable prices.

The broader concern among consumer advocates is that AI accelerates the shift from risk pooling — the traditional mutual insurance model where lower-risk customers cross-subsidise higher-risk ones — to pure risk selection, where insurers cherry-pick the most profitable customers and leave the most vulnerable without affordable options. This tension is central to the FCA’s ongoing consultations on AI use in financial services, and is unlikely to be fully resolved without further regulatory intervention.

Your Legal Rights When an Algorithm Decides Your Premium

UK GDPR Article 22 gives you specific legal rights when automated systems make significant decisions about you. If an insurer uses an AI model to decline your application, substantially increase your premium on renewal, or cancel your policy, you have the right to request human review of that decision. The insurer cannot simply rely on the output of its algorithm without human oversight for decisions that significantly affect you. You also have the right to a meaningful explanation of the factors that influenced the decision.

If you believe an AI-driven insurance decision has discriminated against you on the basis of a protected characteristic under the Equality Act 2010, you can complain to the Financial Ombudsman Service (FOS). You can also raise a concern with the FCA directly through its consumer helpline, or with the ICO if you believe your data has been processed unlawfully. The FOS handled over 8,000 insurance-related complaints in 2024-25, and cases involving algorithmic pricing decisions are a growing proportion of that caseload.

In practice, the most effective first step is to ask your insurer to explain your premium in writing. Under Consumer Duty, they must be able to justify the key factors used in pricing and confirm it represents fair value. If the explanation is inadequate or reveals factors you dispute, escalate to the FOS. Shopping around also remains genuinely effective — different insurers use different models, and the variance in premium offers for the same customer from different providers can still be substantial even in an AI-dominated market.

What This Means for UK Insurance Customers

AI-powered insurance is already shaping the market you buy from in 2026. Telematics policies reward careful drivers with premiums that reflect how you actually drive. Smart home devices are beginning to reduce contents insurance costs for participating households. Health and fitness data from wearables is being trialled in life insurance pricing. These changes can produce real benefits for customers willing to share data and who fall into lower-risk profiles under the insurer’s model.

For all customers, the key is to understand and exercise your rights. You can request human review of algorithmic decisions that significantly affect you. You can demand an explanation of your premium factors. You can complain to the Financial Ombudsman if you believe a pricing decision was unfair, unexplained, or discriminatory. You can choose not to participate in data-sharing schemes — though this may mean foregoing premium discounts available to participating customers.

The FCA’s Consumer Duty regime is one of the most comprehensive consumer protection frameworks for AI in financial services anywhere in the world. The challenge for regulators is ensuring enforcement keeps pace with the speed of technological development in AI underwriting. As a UK insurance customer, you benefit from stronger protections than your counterparts in many other jurisdictions. Knowing what those protections are — and when to invoke them — is increasingly important as more of the decisions that affect your daily financial life are made by algorithms rather than humans.

This article is for educational purposes only and does not constitute financial advice.