Most North American insurers and brokers now use artificial intelligence across core parts of their business, but many still can’t show what it’s worth, according to a survey by Moody’s Ratings.
The rating agency surveyed 58 companies across life, health, P&C, reinsurance and broking. About two-thirds had multiple AI applications in production across core operations, with some broader rollout across the organization. About 34% said they were deploying AI widely throughout the business. The rest had more limited use in selected areas.
Moody’s said AI is mainly acting as a tool to make insurers more efficient rather than disrupting their business models. It expects the effect on credit to be neutral to positive, depending on how well companies execute, the quality of their data and their governance.
Generative AI, such as large language models, copilots and chatbots, was the most widely used, cited by 98% of respondents. Traditional machine learning and predictive models were used by 79%, and agentic or workflow automation by 71%.
Among P&C insurers and reinsurers, operations was the most advanced area, with 75% already using AI in production. More than half had AI in production for underwriting and pricing, and half for both claims and loss adjustment and fraud detection. Catastrophe and exposure management lagged, with just 19% in production. Life insurers focused mainly on customer service and distribution, underwriting and operations.
Brokers were roughly in line with carriers. Of broker respondents, 29% reported broad deployment across the organization, 43% had multiple production uses across core operations, and 28% had more limited use.
The most commonly reported benefits were faster processing and cycle times (74%) and productivity gains or headcount avoidance (76%). P&C insurers also pointed to lower loss severity and leakage, reduced loss adjustment expenses and better combined ratios.
AI typically accounts for 2% to 10% of insurers’ IT budgets, and 13% of respondents said it was more than 10%. P&C insurers are spending more heavily than life insurers. Nearly one in five P&C carriers reported spending above 10% of IT budgets, compared with 7% of life insurers, and 44% of P&C insurers reported spending of 5% to 10%, compared with 3% of life insurers.
Measuring the payoff is harder. Only 41% of respondents said they quantify and track benefits for individual use cases. Another 34% quantify benefits but don’t consistently track them, and 24% said the benefits were mainly qualitative. Moody’s said productivity gains don’t immediately show up in financial results, because companies often have to redesign workflows and run old and new processes side by side for a time. Several insurers have not set a formal timeline for measuring returns.
Moody’s said larger insurers are better placed to absorb the investment, because they have bigger technology budgets and more data, while smaller companies may struggle to keep up.
The biggest obstacles to realizing AI’s benefits were skills and talent shortages (81%), data quality (71%) and integration with legacy systems (62%). Nearly half (47%) cited regulatory uncertainty, which Moody’s linked to differing AI rules across jurisdictions. Most respondents (88%) said their AI governance principles align with regulatory guidance.
Moody’s warned that AI changes insurers’ risk profile. It may reduce risks tied to manual processes, but it increases operational, regulatory and litigation risk, along with dependence on key vendors. The rating agency said advances in frontier AI models could significantly raise cyber risk for insurers and the wider financial system by speeding up how quickly software vulnerabilities can be exploited. It cited an April 2026 example in which Anthropic’s Mythos model found high-severity vulnerabilities in most major operating systems and web browsers. Reliance on third-party models and cloud providers also increases exposure to cyber incidents at critical vendors.
For brokers, the survey cuts two ways. Carriers using AI in underwriting and claims should mean faster quotes and settlements, but brokers whose own systems lag may find it harder to keep up with carriers’ speed and data demands. With smaller players facing greater implementation constraints, the gap between large and small brokerages may widen.
The risk findings also matter for clients. The same AI-driven cyber and vendor risks that Moody’s flags for insurers apply to commercial clients relying on AI tools and cloud providers, which brokers will increasingly need to address in cyber and technology placements.
