Major advances in technology bring about fevered hype, hopes and dreams, venture capital… and dread from insurance carriers. How will we insure against loss from technologies that no one really understands? Cyber risk is a recent example. Now, the insurance industry is scratching its collective head about artificial intelligence (AI). I spoke about this topic with Claire Davey SVP, Product Innovation and Emerging Risk at Relm Insurance. Insuring the unknown is what Claire does all day, so she has a distinct and well-informed perspective on the issue.

Claire Davey SVP, Product Innovation and Emerging Risk at Relm Insurance

According to Claire, AI is moving faster than regulation, and traditional insurers can’t keep up. From algorithmic bias to machine-driven cyber threats, businesses deploying AI face risks no legacy policy can cover. Potentially problematic scenarios abound, but high on the list are risks like an AI chatbot using biased, offensive language, AI software inadvertently ingesting private data and making it part of a general AI algorithm, generative AI (GenAI) engaging in “hallucinations” that create liability, and on and on. As adoption accelerates and compliance tightens, insurance becomes the missing bridge between innovation and investment.

Here are some of Claire’s thoughts on the matter:

Q:          AI is moving faster than regulation, and traditional insurers can’t keep up. Can you expand on that? Where are you seeing the biggest gaps today?

A:           In the EU, there is some clarity regarding regulation and the framework provided. The most concerning complexity arises in the US, where there are separate regulations per state, and industry bodies are seeking to implement their own requirements and frameworks. This makes it confusing and unpredictable for businesses using or developing AI, but it also increases the risk exposure for insurers. Without a consistent regulatory baseline, insurers are forced to navigate a moving target, which makes pricing, reserving, and even determining what is or isn’t covered far more challenging than in traditional lines.

Q:          From an underwriting perspective, how does AI change the very nature of risk compared to traditional cyber or liability exposures?

A:           Additional regulatory exposure is one factor. Another is the lack of mature governance procedures and frameworks within many organizations, which leads to oversight gaps.

On top of that, AI’s ability to develop and produce outcomes in ways that were unintended adds a layer of unpredictability that insurers have to account for.

However, I think there are generally a lot of similarities with how cyber risk emerged as an insurable risk. At first, it seemed quite unquantifiable, but over time, insurers developed the tools and data needed to model it. AI will likely follow a similar path, though the velocity of change is significantly faster.

Q:          Do you think the real challenge is that existing insurance products weren’t built to anticipate AI-driven risks, e.g., cyber policies not accounting for model poisoning or adversarial attacks?

A:           It is more the case that underwriting, pricing, and reserving models were not created with the understanding of exposures driven by AI. The issue is not only about coverage gaps but also about the actuarial foundation insurers rely on to set premiums and reserves; these models were not designed for the scale, speed, and novelty of AI-driven exposures.

Q:          When Relm says it “underwrites AI,” does that mean you’re focused on insuring companies building AI systems, companies adopting AI in their operations, or both?

A:           Both. We underwrite companies that are directly producing AI technologies as well as those that are adopting AI in their operations and therefore facing new risk exposures. For us, underwriting AI means recognizing that the risks can emerge anywhere in the value chain, from the developers building algorithms to the end users applying them in sensitive industries. Our role is to bridge that gap and provide coverage that adapts to both sides of the market.

Q:          Looking ahead, what does the insurance industry need to change, in structure, product design, or mindset, to truly adapt to the AI era?

A:           Monitor AI exposures and constantly tweak coverage, pricing, and reserving models. The pace of AI development means insurers can’t rely on static frameworks – they need to evolve coverage dynamically and work in closer collaboration with regulators, technologists, and businesses deploying AI. It’s also about adopting a mindset that sees insurance not just as a protective layer, but as an enabler of innovation, giving companies the confidence to deploy AI responsibly while knowing they have tailored risk transfer mechanisms in place.

Photo by Pixabay: https://www.pexels.com/photo/blue-bright-lights-373543/