Will Your Insurance Pay When AI Makes a Mistake?
Insurers and brokers are grappling with AI-related losses and policy wording. Why you should brief your broker on specific use cases before expanding AI in the business.
Short answer
Insurance cover for AI-related losses depends on policy wording, the cause of loss, and the facts — not on the assumption that existing professional indemnity, cyber, or liability policies automatically respond. Brief your broker on what each system does, ask for written advice on realistic loss scenarios, and align supplier contracts with the limits of insurance.
AI may be answering your customers, preparing advice, and completing routine business tasks. But when it makes an expensive mistake, who pays? The answer may be less certain than business owners assume.
Reporting by The Wall Street Journal on 16 October 2026 examined an emerging problem in the insurance market. Insurers are grappling with possible losses from AI systems, while existing policies may be unclear about whether particular events are covered. On the same day, insurance broker Marsh warned that sweeping exclusions for AI could increase uncertainty and disputes.
That does not mean every insurance policy excludes AI, or that every AI mistake is uninsured. Coverage depends on the wording, the cause of the loss, and the facts of the claim. The point is that assuming protection without asking is risky.
Consider three ordinary business situations. A customer-service assistant gives incorrect advice and a customer suffers a financial loss. A document assistant discloses confidential material to somebody who should not have received it. An automated tool with access to business accounts makes a payment nobody approved. These are different events. Each may raise different questions under professional indemnity, cyber, liability, or crime-related insurance.
There is another complication: a business may rely on a technology supplier to make good the loss, only to discover that the contract limits the supplier's liability. A company could therefore face a customer claim, an insurance dispute, and a supplier that has promised very little compensation.
What to do
Before extending AI into important operations, tell your insurance broker exactly what the systems do. Do not simply say that the business uses AI. Explain whether it gives advice, handles confidential information, makes decisions, sends communications, or can move money.
Ask for written advice about specific examples of loss. Which policies might respond? Are there AI-specific exclusions, conditions, or sublimits? Do defence costs reduce the available cover? What notification obligations apply when an incident first appears? Check the supplier's contractual promises and financial capacity as well.
In parallel, reduce the chance and size of a loss. Limit access to sensitive information, require approval for payments or consequential decisions, keep reliable activity records, and make it possible to stop a malfunctioning tool. Insurance can soften some financial consequences; it cannot undo a privacy breach or restore a customer's trust.
Sources
- The Wall Street Journal, 16 October 2026 — AI and the insurance market.
- Insurance Journal / Marsh, 16 October 2026 — broad AI exclusions and uncertainty.
- ISO/IEC 42001 explained — International Organization for Standardization.
We document where AI is used, realistic loss scenarios, and proportionate controls — giving management, brokers, and insurers a clearer picture through our gap assessment.