The 2026 Evident AI Index, published today, covers 30 of the largest insurers in North America and Europe. It measures AI maturity across four areas: talent, innovation, leadership, and transparency. The findings are blunt. Early movers are not just ahead. They are pulling further ahead, and the gap is compounding.
Allianz tops the ranking, overtaking AXA which led last year. Zurich made the most dramatic jump, climbing eight places from 12th to 4th. That rise was not accidental. Zurich built a shared AI platform called ZurichIQ, deploying it across underwriting, claims, legal, and service operations rather than running separate experiments in each division. One practical result: a partnership with AI firm Cytora cut manual underwriting triage time from 75 minutes to 15 minutes, and pushed the share of submissions processed automatically from 10% to 95%.
The workforce shift is worth pausing on. The overall insurance headcount fell 2.2% over the past year. At the same time, AI-specialist roles at these 30 firms grew 32%. One in every 50 employees at these insurers now holds an AI-specialist role. Nearly 40% of the indexed firms have appointed a senior executive with explicit responsibility for AI, most of those appointments made in the last 12 months.
Three insurers have crossed a threshold that matters to boards and shareholders: public disclosure of actual financial returns. Manulife reported CA$300 million in AI-generated enterprise value in fiscal 2025 and projects CA$1 billion by 2027. Generali disclosed approximately €100 million in bottom-line impact from AI in fiscal 2025, with a 2027 target above €350 million. These are not projections dressed up as results. They are reported numbers from companies large enough to face serious scrutiny if the figures were inflated.
Why does AI in underwriting matter more than AI in administration? Because claims typically eat between 60% and 80% of every dollar of premium an insurer collects. A small improvement in the accuracy of risk selection, or a meaningful reduction in fraudulent claims, produces a financial effect that administrative savings simply cannot match. That is the logic driving investment toward core risk decisions rather than back-office workflows.
The next shift is already underway. A quarter of newly disclosed AI use cases now show evidence of systems that operate across multiple steps autonomously, initiating actions, checking results, and proceeding without a human approving each move. Six months ago, that share was one in twenty. This is not theoretical. Allianz, for example, built a system called Project Nemo that uses seven coordinated AI agents to handle food spoilage claims, cutting resolution time from days to hours.
For businesses that buy insurance, particularly commercial and specialty lines, this has practical implications. Underwriting decisions are being made faster, using more data, and with less human discretion at each step. That means a stronger data record on your side matters more than it used to. Companies that can provide clean, structured risk information will move through underwriting faster and with more favorable terms. Companies that cannot will face slower processes and less pricing flexibility.
Regulators are moving in parallel. As of early 2026, 23 US states have adopted the insurance regulator's model guidelines on AI use, and New York has required insurers to explain how AI factors into underwriting and pricing decisions. Insurers that built governance processes early are in a better position to meet those requirements. Insurers that did not are facing a simultaneous pressure from regulators and competitors.
The index assessed 30 firms. Twenty-nine of them improved their AI maturity scores year on year. The one that did not is in an increasingly uncomfortable position.