ScienceSoft, a McKinney, Texas-based IT consulting and software development firm, has published research projecting that by 2028, 60% to 80% of new policies and renewals in errors and omissions, directors and officers, employment practices liability, and cyber insurance will factor AI risk into underwriting decisions.
Despite that shift, the firm expects most midsize US insurers to continue covering AI-related exposure through existing policy lines rather than standalone AI-specific products.
The report projects the AI-specific insurance segment will grow from roughly $40 million in 2024 to $4.8 billion by 2032, an annual growth rate near 80%. Even at that pace, ScienceSoft estimated the segment would still account for only about 0.34% of total commercial property and casualty premiums by 2032, a niche by almost any measure despite the rapid percentage growth.
ScienceSoft is not an insurance rating agency, actuarial consultancy, or independent research house. It's a technology vendor that has built underwriting automation systems, claims processing software, and compliance tools for insurance carriers, brokers, and insurtechs since 2012, including, by its own account, a large-scale automated underwriting system for a global commercial carrier with more than $30 billion in assets.
That means the company publishing this forecast about how insurers will need to adapt underwriting for AI risk is the same type of firm insurers would hire to build the underwriting systems capable of doing exactly that.
This doesn't make the analysis wrong, but it's a meaningful detail missing from every version of this release that's circulated across newswire aggregators, and it's the kind of commercial interest that warrants disclosure alongside the findings themselves.
The report's directional claim, that exclusions and adapted policy wording will spread faster than standalone AI products, lines up with a pattern Insurance Business has documented in detail throughout 2026.
The Insurance Services Office published a set of standard generative AI exclusion forms in July 2025, and by August 2026 a nationwide review of state filings found 4,078 records across 49 states adopting versions of those exclusions, with 2,369 already in force across commercial general liability, umbrella, businessowners and E&O policies contractors commonly carry. By April 2026, W.R. Berkley, Chubb, Travelers, Berkshire Hathaway and AIG had each filed to adopt these endorsements or their own proprietary AI exclusion language.
That exclusion trend is fracturing differently across individual lines. Insurance Business has reported that AI exclusions remain in closer to 10% of the EPL market so far, even as litigation like Mobley v. Workday sharpens questions about algorithmic hiring liability, while D&O insurers have so far stopped short of imposing broad AI exclusions, with brokers describing the eventual shift toward exclusions as likely but not yet arrived.
Gallagher's own 2026 AI Adoption and Risk survey, cited in Insurance Business's reporting, found one in five insurance professionals say their insureds have already experienced losses linked to AI risk, while fewer than half of organizations have a formal AI risk management framework in place, the exact silent-risk dynamic ScienceSoft's report describes as driving insurers toward clearer policy language.
HSB has taken the opposite approach in one corner of the market, launching an affirmative AI liability product specifically to write back coverage insurers might otherwise exclude, evidence that a small standalone segment is indeed forming even as the broader market leans on exclusions and endorsements, consistent with ScienceSoft's own framing of AI-specific coverage as a genuine but narrow niche.
ScienceSoft cites a 262% rise in publicly documented AI incidents from 2022 to 2025 as the demand-side driver behind this shift. That figure is broadly consistent with incident tracking maintained by the AI Incident Database, an independent, publicly searchable repository of AI harms that Stanford's annual AI Index has also cited in its own reporting on rising incident volumes over roughly the same period, though ScienceSoft's report doesn't specify which underlying database its own figure draws from.
For carriers and brokers, the report's most practically useful claim, that underwriting scrutiny of AI governance and controls will intensify well ahead of any move toward standalone AI products, is independently supported by Insurance Business's own reporting throughout 2026 on carrier exclusion filings and broker commentary.
But given ScienceSoft's own commercial position building the systems that would perform exactly that kind of governance assessment, brokers and risk managers evaluating this report's specific market-size projections, the $40 million to $4.8 billion growth trajectory and the 0.34% premium share figure in particular, should treat those numbers as one technology vendor's proprietary estimate rather than an independently audited market forecast, and weigh them against the carrier-level filing data and broker commentary Insurance Business has already gathered directly from the market.
