Insurance: when AI dissolves risk pooling

Context

An article published by e27 (July 13, 2026, reposted on July 24) argues that the widespread use of AI in underwriting and pricing dismantles the historical logic of the risk pool in insurance. The thesis: as soon as insurers can price each individual very finely based on behavioral, biometric, and contextual data, mutualization becomes an exception rather than a rule.

The Data

• Source: e27, July 13, 2026.
• Angle: Structural (dissolution of the risk pool), not technical.
• Markets mentioned: health insurance, auto, home.

Analysis

The argument hinges on a simple shift. Historically, insurance is based on three assumptions: limited information about the insured, averaged premium, implicit redistribution of "low risks" to "high risks." AI breaks the first assumption. But if the information becomes fine enough to discriminate individually, then the premium tends toward the expected cost—and the insurance function reduces to a simple advance loan. High-risk profiles no longer find takers; low-risk profiles leave the pool. The mechanism is already at play in health insurance in the United States with wellness programs.

Scenarios

• Base: Regulators (EU, California, NY) impose limits on pricing granularity—return to a constrained risk pool.
• Alternative: Emergence of a two-tier market—private "fine" insurance for good profiles, public safety net for others.
• Pessimistic: Systemic drift—entire segments uninsurable, cost transfer to the public sphere without fiscal compensation.

Implications for the Professional

For an insurance executive, the challenge is no longer the power of the model but its regulatory framework. For a regulator, it's time to anticipate a reform of pricing rules (transparency, bias tests, granularity thresholds). For a professional underwriter, the AI-individualized premium is coming to companies via benefits—expect to see employee behavioral scores turn into variables for collective health insurance.

Signals to Watch

• Decisions in Colorado/California on actuarial algorithms.
• Position of ACPR/EIOPA on the use of AI in life/health pricing.
• First US class actions on algorithmic discrimination in insurance.

Our Take

AI is not just a tool for insurance—it's a force that puts the social contract that underpins it under tension. The debate will quickly shift from the technical to the political. Insurers who anticipate this will keep the upper hand.