"Right now, a lot of clients think if they have cyber insurance, they have AI coverage, and that's a very common misconception," said Jennifer Wilson, senior director, cyber at WTW, at a recent Insurance Business cyber roundtable. "AI is a technology. AI is the tool. It's not the peril. So AI could cause a loss under cyber. It could create a loss under tech E&O or D&O or EPL."
That distinction determines which policy responds when something goes wrong - and the answer is rarely as simple as "the cyber policy."
Joe Earle, cyber group practice leader at C3 Risk and Insurance Services, described two misconceptions that run in opposite directions.
"One of the biggest misconceptions is if you don't have AI affirmative language, that you don't have AI coverage. A lot of companies still cover AI under their software language. If you're developing software and that software includes AI, it's still covered under software development, part of it on a technology errors and omission policy."
No affirmative language does not always mean no coverage.
"Another misconception is that if you have AI affirmative coverage, you have coverage for all AI. There are differences between types of AI. There's also differences between the way you use AI, whether it's your employees using the AI through an LLM, and also if you have AI agents out there, with or without supervision, making decisions on your behalf."
Affirmative language does not guarantee comprehensive coverage either.
Wilson added the carrier-level dimension that brokers need to be navigating. "There are carriers out there that are promoting that they have AI coverage, and it's misleading to policyholders thinking that they have the coverage. They're just slapping AI language, say, on their cyber policy, but they don't have it on the tech E&O."
She went further. "There are some carriers out there who do have affirmative AI, and it's narrowing the coverage, not broadening it. And then there are other carriers that truly have effective and appropriate affirmative AI language that is broadening the coverage. And without that broadened coverage, there is no coverage on the standard cyber and tech E&O policy."
Three carrier positions - misleading, narrowing, genuinely broadening - and the difference between them is not visible from the marketing language.
A Stanford-affiliated advertiser used AI to generate an image for a cafeteria banner. The image included a photo of three students. The AI altered the race of one of them. The result potentially touched EPL, D&O, media liability, and cyber simultaneously.
"The one policy that we never really talk about that kind of gets added on sometimes to a cyber policy is a media liability policy," Earle said. "People are using AI for media more and more often. We're opting out of using a graphic designer, and we're skipping some approvals at the higher levels, and we're just putting stuff out there."
Organizations using AI for content generation are routinely bypassing review steps that previously existed. When something goes wrong, the question of which policy was designed to respond is not always answered by the one that seems most obvious.
Keith Savino, CEO US at Emergence Insurance, identified the exposure that lands closest to home.
"Your insured is maturing every day of the week, and they're getting smarter and smarter. And by the way, even if they don't know the answer, guess where they're turning to ask the question? To AI.
So they're asking AI to compare three policies. They're coming to you with a list of questions. And then if you're using the same AI to answer those questions, and both sets of AI are hallucinating with each other, that might be called an E&O exposure."
Wilson put the cost of the knowledge gap in concrete terms. A client came to her team with a competing quote $50,000 cheaper than what WTW was presenting. "I had to spend time communicating back and forth with the general counsel, helping her to understand you truly get what you pay for. And there were so many critical coverages that were not captured in the much cheaper quote. And these are coverages that a sophisticated client was not aware of."
A sophisticated buyer, not a naive one, unable to distinguish coverage that looks similar on price but differs materially on substance. That is the gap brokers need to be closing - because the alternative, Wilson noted, is clients "sitting in the seat of making investment choices with respect to insurance and not really understanding the fine print or the lack thereof."
Savino's close was direct. "You need to know what's in there, and you need to know what you're selling."
