AI Errors Insured by Munich

What happened

Munich Re and other insurers have introduced new coverage for US mortgage lenders, insuring against errors from AI screening systems. This coverage addresses risks including AI underperformance, model failure, and discrimination within AI models used for borrower risk assessment and fraud detection. It protects against contractual, financial, and legal liabilities. Premiums are determined by the AI model's robustness, quantifying the probability and severity of potential underperformance, enabling lenders to deploy AI solutions with financial protection against model underperformance.

Why it matters

The introduction of AI error insurance shifts the burden of AI model failure risk from direct operational loss to an insurance cost, potentially reducing internal capital requirements for lenders. However, it introduces a new dependency on third-party risk assessment and premium structures, requiring increased due diligence from procurement and risk management teams to evaluate policy terms and the insurer's model robustness quantification. This creates an oversight burden for compliance and IT security to ensure AI model changes do not invalidate coverage or alter risk profiles.