Insurers say they’ll exclude AI risk first; Allianz says AI losses are already being paid under existing policies. Both are right: why, which side your policy is on, and three steps — check the policies, ask your broker, list your AI use.
Insurers say: AI risk — exclude it first, don’t cover it. Allianz says: AI losses have already been paid under existing policies — just never tagged as AI losses. The two statements contradict each other, and both are right. The question is: which side is your policy on?
AI-caused losses are already happening. AI says the wrong thing to a customer, generates bad information, an algorithm’s decision causes harm — and those losses get paid under general liability, cyber and professional liability policies. Nobody tags them as “AI losses” when the claim is filed.
Untagged means insurers can’t measure how much AI risk they’re actually carrying. And when insurers can’t measure a risk, their instinct is to exclude it. A CHES executive said it plainly: insurers would rather write AI into the exclusions than cover it.
So here we are: losses are already being paid, insurers are preparing to close the door, and the policy you think covers you may not say so.
You think: I bought a policy, so if AI causes a problem, insurance has my back. Reality: most standard policies never mention AI. If it isn’t written down, it isn’t covered.
You think: they’re paying these losses now, so they’ll keep paying. Reality: they’re paying now because the wording hasn’t changed and the losses haven’t been identified. Once insurers can measure them, they will either exclude AI or price it separately. One renewal with new wording, and your AI exposure is bare.
You think: AI is a small thing, it can’t cause big losses. Reality: AI runs on shared models. One model failure can trigger claims from a large number of customers at once — exactly the systemic risk insurers fear most, and most want to exclude first.
A decade ago, cyber losses were hiding inside property and liability policies — untagged, unpriced, quietly paid. When insurers finally measured them, the industry either excluded cyber or moved it into standalone, separately priced cyber policies. That was silent cyber.
Now it’s AI’s turn, and the path will likely be the same: AI exclusions and AI-use disclosure requirements at application first, then standalone AI liability products and endorsements. The difference is, this time you can act before the door closes.
Not “review your policy” in the abstract. Three steps.
Step one: check three policies. Cyber, errors and omissions (E&O), general liability. In each one, look for three terms: AI, automated decision, algorithm. If you can’t find them, your AI activities are not within scope.
Step two: ask your broker three questions. Are my AI activities — customer service, content generation, data analysis, automated decisions — covered? Will my renewal add an AI exclusion? If AI isn’t covered, can an endorsement add it, or is there a standalone AI liability policy?
Step three: list your AI use. Write down what your business uses AI for, and give the list to your insurer. Insurers may start requiring disclosure; disclosing early beats being told after a loss that you never mentioned it.
One side says they won’t pay. The other side is already paying. Which side your policy falls on depends on whether AI is written into the wording. If it isn’t written down, it isn’t covered. Before an exclusion lands on your policy, get the three steps done.
Such losses are already being paid under general liability, cyber and professional liability policies; they just aren’t tagged as AI losses when the claim is filed. But most standard policies never mention AI, and if it isn’t written down it isn’t covered. One renewal with new wording and your AI exposure can be bare.
About a decade ago, cyber losses were hiding inside property and liability policies — untagged, unpriced, quietly paid. Once insurers measured them, the industry either excluded cyber or moved it into standalone, separately priced policies. AI risk is likely to follow the same path.
Three questions: Are my AI activities — customer service, content generation, data analysis, automated decisions — covered? Will my renewal add an AI exclusion? If AI isn’t covered, can an endorsement add it, or is there a standalone AI liability policy?
2 emails a week: rate moves, regulation and money-saving tips. Unsubscribe anytime.
You are subscribed right away — we send a short notice you can unsubscribe from at any time.