Drawn from published evidence and regulatory guidance specific to insurance. Each is pre-scored on a 5×5 likelihood × impact matrix in the Risk Register tool and referenced in the generated policy.
CriticalLikelihood 4 · Impact 5
AI Underwriting and Pricing Encoding Protected Characteristic Proxies at Actuarial Scale
AI insurance pricing and underwriting models trained on historical claims, customer, and socioeconomic data encode proxy variables correlated with race, ethnicity, disability, sex, religion, and age — including postal code, occupation, credit score, and health service utilisation patterns — producing systematically less favourable insurance terms, higher premiums, or coverage refusals for protected characteristic groups at a granularity and scale that obscures discriminatory patterns from routine actuarial review and regulatory examination.
HighLikelihood 4 · Impact 4
AI Claims Fraud Detection False Positives Causing Wrongful Denial to Legitimate Policyholders
AI fraud detection and claims assessment systems produce elevated false positive fraud flags for legitimate claims from certain policyholder groups — including elderly claimants, disabled claimants, and claimants from minority ethnic backgrounds — resulting in wrongful claim denials, prolonged investigation causing financial distress, and reputational harm to policyholders subjected to unfounded fraud allegations by AI-driven claims systems.
CriticalLikelihood 3 · Impact 5
AI Solvency Model Failure Causing Inadequate Capital Reserves or Catastrophic Loss Underestimation
AI actuarial models used in Solvency II capital calculation, catastrophe loss modelling, life reserving, or reinsurance pricing produce materially inaccurate outputs — through model drift as climate risk and mortality patterns shift beyond training distributions, or through AI confidence exceeding genuine predictive capability — causing insurers to hold inadequate capital buffers, undercharge reinsurance premiums, or misestimate exposures that crystallise as solvency-threatening losses.
HighLikelihood 4 · Impact 4
AI Price Walking and Loyalty Penalty Systematically Harming Existing Policyholders
AI dynamic pricing and renewal optimisation systems identify policyholders unlikely to shop around — through AI analysis of renewal behaviour, digital engagement, price sensitivity signals, and loyalty indicators — and systematically price renewals above new customer acquisition rates for equivalent coverage, exploiting policyholder inertia at scale with particularly severe harm to elderly, vulnerable, and digitally disadvantaged customers least likely to comparison shop.
CriticalLikelihood 3 · Impact 5
AI Health Data Inference in Life and Health Insurance Creating Unlawful Discrimination
AI life insurance underwriting and health insurance pricing systems infer health conditions, mental health status, genetic risk factors, and disability status from non-medical data — including prescription purchase patterns, fitness tracker data, and postcode-level health statistics — using these inferences to adjust premiums, impose exclusions, or decline coverage without disclosure to the applicant, violating GDPR Article 9, GINA in the US, and the Equality Act 2010 in the UK.
HighLikelihood 4 · Impact 4
Generative AI in Claims Documentation Enabling Fraud or Causing Wrongful Denials
Insurers or claimants use generative AI tools to produce synthetic claims documentation — including AI-generated medical reports, AI-fabricated damage photographs, AI-written repair estimates, and AI-authored expert opinions — submitted as genuine evidence in insurance claims processes, causing fraudulent payments where AI-generated evidence supports false claims or causing wrongful denials where AI-generated insurer documentation misrepresents the factual basis for repudiation.