Insurers race to cover AI errors

Summary

This Marketplace Tech episode examines insurance as enabling infrastructure for business AI adoption. It contrasts Corgi’s explicit AI liability coverage with Claimy’s use of agent testing to expose failures and adjust premiums, while Gallagher explains why incumbent insurers are considering standardized wording and possible AI exclusions.

The central constraint is AI Insurance Data Scarcity: insurers have historical loss models for familiar hazards but little comparable evidence for rapidly changing AI systems. The episode therefore presents insurance as both a financial backstop and an incentive for pre-loss testing, without establishing mature pricing, settled liability boundaries, or a claims record.

Key Claims

• More than one-fifth of U.S. businesses use AI in daily operations, according to a Goldman Sachs figure cited by the episode, creating a growing set of operational and liability exposures.
• A Deloitte forecast cited by the episode places the global AI-insurance market near $5 billion by 2032, but the program provides no underlying forecast methodology.
• Corgi says its AI liability product explicitly covers harms from prohibited statements, incorrect calculations, and service outages caused by AI systems.
• John Farley says traditional insurance is moving from silence about AI toward scrutiny, standardized wording, regulatory review, and possible exclusions.
• Unlike catastrophe insurance, AI coverage lacks a long loss history and stable modeling tools, making exposure and premium estimates difficult.
• Claimy tests customer agents for claim-producing failures such as disclosure of sensitive customer data, allows remediation and retesting, and links better test scores to lower premiums.
• Connecting test results to price can turn insurance from post-loss payment into a continuing risk-management incentive.

Key Quotes

“from silence to scrutiny” - John Farley’s description of the insurance market’s changing treatment of AI.

“There wouldn’t be skyscrapers in New York if it wasn’t for insurance.” - the analogy used by Claimy’s founder to frame coverage as infrastructure for innovation.

Connections

• Marketplace Tech - show context.
• Corgi, Nico Lacqua, and AI Liability Insurance - explicit coverage for AI-caused business harm.
• Claimy, Ines Butemacha, and AI Agent Risk Testing - failure testing, remediation, rescoring, and premium incentives.
• Gallagher, John Farley, and AI Insurance Data Scarcity - incumbent-market scrutiny, exclusions, and modeling limits.
• Goldman Sachs and Deloitte - adoption and market-size figures cited by the episode.
• Insurance Risk Transfer and AI Governance And Compliance - broader risk-transfer and enterprise-control context.

Contradictions

• No settled contradiction is identified. The episode records a market split between explicit AI coverage and possible AI exclusions, but these are different insurer responses to uncertainty rather than incompatible factual claims.
• Adoption, market-size, policy-coverage, and premium claims remain source-scoped because the short episode does not provide underlying datasets, complete policy language, prices, regulatory approvals, or actual claims outcomes.