Specialist AI insurance is already on the market, but it may take real business losses before companies start buying it in earnest, according to Neil Kessler (pictured), chief executive officer, specialty and benefits, at CRC Group.
Modeling AI insurance on cyber’s growth between 2009 and 2017, the Deloitte Center for Financial Services projected that insurers could write about $4.7 billion in annual global AI insurance premiums by 2032, a compound annual growth rate of around 80%.
But CRC, which places AI coverage and is developing its own solutions through its underwriting business, has found that having products to offer has not produced a rush of buyers.
“There are AI insurance products that exist today that we are selling. Are those flying off the shelves?” Kessler told Insurance Business, “No, but we do have those as insurance products that exist, and we are able to offer those coverages.”
Part of the problem, he said, is that little work has been done to put a price on what an AI-related loss could actually cost. Kessler sees an opportunity for excess and surplus lines specialists to help answer that question.
The challenge starts with understanding the potential damage. Kessler said he had seen relatively little work, from his own perspective, on the economic losses businesses could face from AI risk, which makes the conversation about insurance spending harder. An acknowledgment that something could go wrong still leaves business owners weighing the potential cost against the value of buying protection.
“Everyone knows that there’s a risk there,” Kessler acknowledged. “I think the question is coming down to: what is that risk from an economic perspective? And what is the value judgment of how I cover that risk?”
Before businesses commit budget to the coverage, those risks “have to become more tangible.” The CRC leader expects actual business losses to make that happen, drawing a comparison with the development of cyber insurance and employment practices liability insurance.
“It’s kind of like the early days of cyber or the early days of EPLI. You’re going to have to have business losses that result in the need for boards or business owners to sit down and decide that they need to allocate funds to buy insurance coverage for this, to cover that risk,” Kessler said.
His assessment puts loss scenarios at the center of the placement discussion: what could happen, what the financial consequences might be and whether available coverage addresses the exposure.
Larger and smaller businesses will likely view the risk differently, as federal data show a clear divide in adoption. Between December 2025 and May 2026, 17% to 20% of US businesses reported using AI in the previous two weeks, according to the Census Bureau’s Business Trends and Outlook Survey.
The rate was 37% among firms with at least 250 employees and under 20% among firms with four or fewer. Use rose among firms with 20 or more employees but did not change significantly among smaller ones.
Alongside the question of potential losses sits another problem for retail and wholesale brokers: understanding how AI exposures interact with insurance clients already buy.
Kessler pointed to the relationship between cyber, technology and AI coverage, including what falls within a general liability policy. He described questions about what standard markets are saying, and what coverage is available in the nonstandard market, as evolving weekly.
“There are different components to this that are going to have to be figured out by insurance carriers, by retail insurance agents, by wholesale brokers,” Kessler said. “Over the next six, 12, (or) 24 months, we’ll start to sort some of this stuff out.”
Standard forms are already moving. ISO, Verisk’s standard-forms division, has issued generative AI exclusion endorsements that became available to carriers in January 2026. The broadest, CG 40 47, removes CGL coverage for bodily injury, property damage and personal and advertising injury arising from generative AI. CG 40 48 applies only to personal and advertising injury, and CG 35 08 addresses products-completed operations
Read more: ISO's generative AI exclusion is already on thousands of CGL policies
Kessler noted that some carriers have adopted ISO’s approach in the admitted market. W.R. Berkley, Chubb, Travelers, Berkshire Hathaway and AIG had each filed to adopt the ISO forms or proprietary equivalents by April, and Berkley has introduced an absolute AI exclusion across its D&O, E&O and fiduciary products, according to the same report.
A review of state filings by Trades Coverage, a licensed insurance producer and marketplace, found 4,078 generative AI exclusion records across 49 states and Washington, DC, through July 31, 2026, of which 2,369 were in force. The review, which covered admitted-market filings, found no admitted standalone product for contractors that replaces the excluded coverage.
Read more: The AI clause that may already be on your contractor client's policy
For its part, CRC Specialty has set up a center of excellence and working group on the issue. Kessler said specialists are helping retailers understand inclusions and exclusions, while CRC’s underwriting business is working on coverage solutions. He also pointed to MGAs established to offer AI-specific protection.
With more product innovation likely in the short term, the immediate task for brokers is connecting what carriers are prepared to offer with the protection retailers are seeking.
“You’re now seeing carriers come out with positions and retailers come out with expectations,” Kessler said. “Our job is to sit in the middle of those and try to bridge those gaps and provide solutions where our clients want them.”
