Accelerate enterprise AI adoption with an end-to-end risk management platform

Not reliably. Marsh finds generative-AI exposure runs across virtually all lines of commercial insurance — cyber, tech E&O, media, D&O, employment practices, IP, general and product liability — not just cyber. And as Insurance Thought Leadership puts it, many of those AI risks are “sitting silently inside existing policies, often unpriced, unmanaged, and waiting to materialize,” because AI losses don't map cleanly onto traditional lines.

Source: Marsh · Insurance Thought Leadership

It is already producing claims. Fact.MR projects AI-related legal claims worldwide to exceed 2,000 by the end of 2026, and identifies hallucinations and errors as the single largest driver — roughly 31% of the emerging market — alongside rogue autonomous actions and bias in lending and hiring. The Geneva Association describes the core failure mode plainly: models that “confidently output false or misleading information” or inadvertently replicate copyrighted content.

Source: Fact.MR · The Geneva Association

It is coverage that was never affirmatively written for artificial intelligence — the form simply didn't contemplate it. Insurance Thought Leadership describes these as risks “sitting silently inside existing policies, often unpriced, unmanaged, and waiting to materialize.” Whether such a policy responds is then decided at claim time rather than before it, which is precisely the ambiguity an assessment is meant to remove.

Source: Insurance Thought Leadership

Partly, and unevenly — which is the point of assessing it. The Geneva Association flags three insurability obstacles: an excessive maximum possible loss from widespread failures, large average losses from events like misinformation or regulatory fines, and severe information asymmetry over how a business actually governs its AI. The market is responding three ways at once: AI endorsements bolted onto cyber and E&O, new underwriting (including parametric triggers), and standalone products such as Munich Re's aiSure.

Source: The Geneva Association

Through technical translation between how a system works and how a policy is worded. Fact.MR reports that specialist brokers and MGAs already hold roughly 46% of this market on the strength of that underwriting expertise, and the Geneva Association notes insurers increasingly approve cover only after “scrutinising insureds' AI systems and governance practices.” A broker that doesn't issue its own paper can run that scrutiny on your behalf and place across markets.

Source: Fact.MR · The Geneva Association

It is the lever the research keeps returning to. Insurance Thought Leadership argues that robust governance, transparent validation, and continuous monitoring both reduce insurance costs and demonstrate trustworthiness to underwriters — treating “the whole workflow from data pipelines to prompts and APIs as the governed unit.” The Geneva Association expects insurers to require human oversight, bias checks, model audits, and reporting as conditions of cover. Our governance score is built from exactly those controls.

Source: Insurance Thought Leadership · The Geneva Association