Insurance Pricing Practices and Automated Underwriting

Insurance pricing practices have traditionally relied on actuarial models and historical data to assess risk and set premiums. In recent years, financial institutions have increasingly adopted automated underwriting systems to streamline these processes. These systems use machine learning algorithms and data analytics to evaluate applicants quickly and consistently. The shift towards automation raises important compliance considerations under existing UK regulatory frameworks, particularly regarding fairness, transparency, and accountability.

Automated Underwriting and Risk Assessment

Automated underwriting systems process large volumes of data to make decisions about insurance eligibility and pricing. These systems often use factors such as credit scores, demographic data, past claims history, and behavioural indicators. For example, a motor insurer might use telematics data from a driver’s smartphone to adjust premiums based on actual driving patterns. While such systems can improve efficiency, they must not introduce bias or discrimination. The Financial Conduct Authority (FCA) expects firms to ensure that automated decisions are fair and compliant with anti-discrimination laws.

• Systems must not rely on protected characteristics such as age, gender, or ethnicity unless directly relevant to risk assessment.
• Algorithms must be tested for bias using diverse datasets to prevent unintentional discrimination.
• Decisions must be explainable, especially where they affect consumers negatively.

Compliance Considerations Under UK AI Regulations

As of 2 February 2025, the EU AI Act’s Article 5 prohibition on certain AI practices applies in the UK. This includes bans on AI systems that manipulate human behaviour or use biometric categorisation for determining sensitive attributes. Insurance firms using AI for pricing must ensure these systems do not fall into these prohibited categories. The FCA expects firms to apply AI governance frameworks that align with these obligations. The Digital Omnibus on AI, which came into force on 27 July 2026, further clarifies how these rules apply to financial services.

Under the AI Act, firms must implement AI governance structures that meet the requirements of ISO/IEC 42001:2023. This standard provides a framework for managing AI systems through processes such as risk assessment, data governance, and audit trails. The standard requires firms to maintain records of how AI models are developed, tested, and monitored. For example, an insurer using machine learning to price life insurance must document how data was selected, how models were validated, and how outcomes are reviewed.

• ISO/IEC 42001:2023 Clause 7.2 requires firms to identify and assess AI-related risks.
• Clause 8.3 addresses data quality and integrity, which is essential for fair pricing.
• ISO/IEC 42006:2025 governs certification bodies, ensuring that firms can obtain proper validation of their AI systems.

Transparency and Explainability

As of 2 August 2026, Article 50 of the EU AI Act requires transparency in AI systems. Consumers must be informed when they are interacting with an AI system, particularly if it affects their rights or interests. In insurance, this means that applicants must know if their pricing is derived from an automated system. Firms must also provide explanations for decisions, especially when they result in higher premiums or rejection of coverage. The FCA’s approach to AI governance encourages firms to develop explainable AI models that can be reviewed by compliance officers or auditors.

For example, an insurer using AI to price home insurance must be able to explain why a particular applicant was assigned a higher premium. This may involve providing information such as the data points used, the weight given to each factor, or the model’s confidence level. The ability to explain decisions is not only a regulatory requirement but also a way to build consumer trust. Firms must also ensure that any AI-generated pricing decisions are reviewed by human specialists to prevent systemic errors or bias.

As of 2 December 2026, generative AI systems placed on the market before that date must meet machine-readable marking requirements. This is relevant for insurers using AI tools to generate pricing models or reports. The marking must indicate that the output was generated by AI, which helps maintain transparency and allows for proper oversight. Firms must also ensure that any AI models used for pricing have been properly tested and validated before deployment. The FCA expects these models to be reviewed regularly to ensure they continue to meet regulatory standards.

By aligning AI practices with existing UK regulatory frameworks, firms can reduce compliance risk while maintaining the benefits of automation. The focus must always be on fairness, transparency, and accountability. The evolving AI landscape requires continuous attention to governance, data quality, and human oversight. Firms must stay informed of regulatory developments and ensure that their AI systems meet both current and future compliance expectations.