Property & Casualty: What Risk Managers Should Know About AI
AI is reshaping how your insurance gets priced, how your claims get handled, and what your business is exposed to. A briefing for risk managers and CFOs on what to do about it.
AI and generative AI adoption in insurance underwriting is projected to jump from 14% today to 70% within three years, according to Accenture's survey of more than 400 underwriting executives. About 77% of P&C carriers have already adopted AI in some form, and leading insurers are cutting underwriting costs by up to 40% through automation, per McKinsey.
That means your next placement may be evaluated by an AI-driven underwriter before a human ever sees it. The implications reach every policy you buy, every claim you file, and increasingly, every AI tool your own business uses.
Where AI Is Helping
The upside is real. Clean, well-structured submissions are moving through markets faster than ever, and a well-told risk story matters more than it used to. AI is surfacing favorable loss patterns on well-run accounts that manual review can miss, particularly long-tenured books with quiet histories. Routine claims are getting straight-through processing, which moves cash back to your business faster. And AI-driven fraud detection is gradually compressing premium for clean books as carriers identify organized and opportunistic fraud that used to spread across the pool.
Your Risk Profile Is Shifting
The other side of the ledger gets less attention but matters more. Your own use of AI in customer service, HR, product design, supply chain, and marketing is creating exposures across multiple lines: E&O when an AI tool gives a customer wrong information, D&O when shareholders question oversight of AI-related decisions, Cyber when AI vendors expand your attack surface, EPL when AI-driven hiring or performance tools encode bias, and Product Liability when AI features embedded in your products cause harm.
Most of these exposures can be present whether or not your business has explicitly “adopted AI.” If your vendors have added AI capability to existing tools, and many have, those exposures came along with them.
“Silent AI” and the Coverage Question
Coverage for AI-related claims is being fought over in real time. Most policies were not written with AI in mind. What is covered, what is excluded, and what falls into the “silent AI” gap is being argued among carriers, brokers, and policyholders right now. Treating this as a paperwork exercise tends to surface problems at claim time rather than at placement, which is the wrong order.
Regulators are moving in the same direction. The Colorado AI Act, NYC Local Law 144, the EU AI Act, and a fast-growing patchwork of state laws are creating compliance obligations that were not on the radar two years ago. Compliance and insurance now need to talk to each other on this topic in a way they did not before.
When the Answer Is “The Model Said So”
A practical wrinkle: AI-driven underwriting can be opaque. A declination or steep rate hike from a model may come with limited explanation and limited appeal. The same is true on the claims side, where AI is increasingly involved in initial coverage and adjudication decisions. Knowing how to engage when the answer is “the model said so” is part of the work now.
There is also a vendor concentration problem. Heavy reliance on a small number of AI vendors for critical business functions introduces business interruption and contingent cyber risk that traditional analyses tend to underweight.
Questions Worth Asking Now
Of your carriers: How are you using AI in pricing and claims on our account, and what data feeds those models? Of your internal teams: Where is AI embedded in our operations, what contracts do we have with AI vendors, and is our governance adequate? Of your general counsel: How are AI-related claims and disputes showing up in our sector?
The Artemis Approach
AI sits inside our Client Management Process the same way every other risk does. Strategy is where we map your AI exposures, both inside-out (your operations) and outside-in (carrier and market behavior). Execution is where AI-aware submissions, coverage gap analysis, and AI-informed claims advocacy happen across the year, not just at renewal. Results is where we measure outcomes, document decisions, and reset for the next cycle with what we learned.
Your program needs a broker paying attention to all three: how your risk gets priced, how your claims get handled, and what your business is exposed to. We are.
Next in the series: Employee Benefits.
