Insurance Carriers Quietly Back Away From Covering AI Outputs

Why It Matters

The retreat limits companies’ ability to insure AI‑driven disruptions, raising operational risk and cost. It also accelerates the push for transparent AI governance as a prerequisite for coverage.

Key Takeaways

• Carriers are declining AI‑related cyber and E&O coverage.
• Some insurers raise premiums for AI risk exposure.
• Exclusions target AI vendors more than AI‑using businesses.
• Underwriters demand detailed AI governance documentation.
• Governed AI may stay insurable; experimental AI faces refusals.

Pulse Analysis

Insurance underwriters are entering a period of heightened caution as AI becomes a core business tool. After early 2025 filings by AIG, Great American and W.R. Berkley to seek regulatory permission for AI exclusions, a wave of carriers has moved from pre‑emptive paperwork to concrete policy language that carves out AI‑generated losses. The primary concern is the “black‑box” nature of many generative models, which makes it difficult for insurers to assess liability, especially in cyber‑risk and errors‑and‑omissions lines. This shift signals a broader industry reassessment of how emerging technologies fit within traditional risk appetites.

For enterprises, the immediate impact is twofold: higher insurance costs and tighter underwriting scrutiny. Insurers are not only refusing coverage for certain AI deployments but also imposing steep premium increases for those they will insure, particularly when the AI is deemed experimental or lacks robust governance. Vendors that sell AI platforms face the harshest treatment, often receiving outright declinations, while end‑users may still obtain coverage if they can demonstrate controlled, governed AI processes. Companies are now required to disclose detailed AI policies, data‑handling procedures, and rollback mechanisms, turning governance documentation into a de‑facto prerequisite for any AI‑related insurance.

Looking ahead, the market is likely to coalesce around standardized AI risk frameworks that satisfy both regulators and insurers. As insurers gather more data on AI‑related claims, they may develop actuarial models that differentiate between “governed” AI—bounded, auditable, and with clear decision pathways—and “autonomous” AI that operates without human oversight. In the meantime, business leaders should prioritize transparent AI governance, conduct regular model audits, and engage insurers early in the development cycle to avoid surprise exclusions. Proactive risk management will not only lower premiums but also ensure that AI continues to drive value without leaving firms exposed to uninsured liabilities.