For decades, the insurance industry has been built on a vast administrative infrastructure: the clerks who process policies, the assistants who schedule adjusters, the data-entry operators who feed information into systems, the claims processors who move paper through pipelines. These workers - overwhelmingly female, often based in smaller cities and college towns far from the coastal tech hubs - have long formed the quiet backbone of the business.
Now, a sweeping new analysis from the Brookings Institution and the National Bureau of Economic Research is putting numbers to what many in the industry have begun to sense: those workers are among the most exposed to artificial intelligence-driven displacement in the entire US economy - and among the least equipped to weather it.
The research, by Sam Manning and Tomás Aguirre of the Centre for the Governance of AI, along with Brookings senior fellow Mark Muro, does something previous analyses have largely failed to do. It does not simply measure which jobs AI can theoretically perform. It asks which workers, if displaced, would have the hardest time recovering - factoring in their savings, age, the density of local job markets, and the transferability of their skills. The resulting picture is sobering for anyone who oversees, employs, or advises workers in the insurance sector.
The warning signs are not theoretical. They are showing up in hiring data right now.
A Q1 2026 Insurance Labor Market Study conducted by The Jacobson Group and Aon's Strategy and Technology Group found that job openings in finance and insurance fell to their lowest monthly level in a decade by December 2025 — dropping from an annual average of 281,000 openings to roughly 138,000 in a single month. The same study found that 43% of insurance industry respondents plan to hold staffing steady over the next 12 months, a figure that rose 10 percentage points in just one year. Automation improvements requiring fewer staff were the most common reason cited by companies that reduced headcount.
The study also found that involuntary turnover across the insurance industry rose 0.6 percentage points year over year — attributed in part to technology advancements and merger and acquisition activity. Meanwhile, P/C industry headcount grew by only 0.81% from January 2025 to January 2026, significantly below the anticipated rate of 1.42%.
Jeff Blair, senior vice president of executive search and business development at The Jacobson Group, said that roles in financial reporting, data synthesis, and aggregation are among those most likely to be displaced by AI, and that call centers, data entry, and transactional operations work face some of the greatest displacement risk. The roles that are growing, by contrast, are experienced underwriters, compliance specialists, analytics professionals, and technologists - positions that require judgment, not just processing.
The industry is, in a phrase, automating from the bottom up.
The new Brookings research quantifies the risk in stark terms. Of the 37.1 million US workers in the top quartile of occupational AI exposure, about 26.5 million also have above-median adaptive capacity — they are, in other words, reasonably well positioned to find new work if displaced. But some 6.1 million workers, representing about 4.2% of the workforce studied, face both high AI exposure and low adaptive capacity. These are the workers with the least runway.
The occupations clustered in that vulnerable quadrant read like an org chart of a mid-sized insurance carrier: office clerks (2.5 million workers), secretaries and administrative assistants (1.7 million), receptionists and information clerks (965,000), medical secretaries and administrative assistants (831,000), insurance sales agents (469,000), insurance claims and policy processing clerks (229,000), and legal secretaries and administrative assistants (155,000).
Adaptive capacity, as the researchers define it, is shaped by four factors: liquid savings, the transferability of skills, the density of the local job market, and age. Workers in these clerical and administrative occupations tend to score poorly on all four. Their savings are often modest, their skills are narrowly applicable, they are more likely to be older workers, and they are disproportionately concentrated in smaller metropolitan areas - university towns, state capitals, and midsized markets in the Mountain West and Midwest - where alternative employment opportunities are thinner.
Geographically, the share of highly exposed but low-adaptive-capacity workers ranges from 2.4% to 6.9% across U.S. metro areas, with a national average of 3.9%. The concentration is highest in places like Laramie, Wyoming; Springfield, Illinois; Carson City, Nevada; and Frankfort, Kentucky - not New York or San Francisco.
If the Brookings research lands with particular weight in the insurance industry, it is in part because of one number embedded in its findings: approximately 86% of the 6.1 million workers identified as facing both high AI exposure and low adaptive capacity are women.
That figure reflects a structural reality that has been building for years. Women dominate the administrative and clerical roles that are most susceptible to large language model automation. Court, municipal, and license clerks are 85% female. Payroll and timekeeping clerks are 89% female. Secretaries and administrative assistants are 96% female. Insurance claims and policy processing clerks are 84% female. Receptionists and information clerks are 92% female.
The gender dimension extends beyond occupational concentration. A May 2025 report by the International Labour Organization and Poland's NASK Research Institute found that if the jobs most highly exposed to generative AI were to disappear, two women would be displaced for every man. A broader analysis cited by the Gender Snapshot 2025 report found that employed women are nearly twice as likely as men to work in jobs at high risk of automation - 4.7% of women's jobs compared to 2.4% of men's, representing approximately 65 million jobs for women globally versus 51 million for men.
In high-income countries, the disparity is more pronounced still. In Australia and New Zealand, for instance, 9.6% of women's jobs are at high risk of automation, compared to 3.5% of men's.
Compounding this exposure is a widening adoption gap. Research by Harvard Business School Associate Professor Rembrand Koning found that women are adopting AI tools at roughly 25% lower rates than men on average. A 2024 survey by the Federal Reserve Bank of New York found that half of men used generative AI tools in the previous 12 months, compared with about a third of women. The reasons are multiple: ethical concerns about the technology, fear of being judged for relying on AI-generated work, and historically lower exposure to STEM fields. The risk is that women in AI-exposed roles who do not adopt AI tools to augment their output may be displaced sooner than those who do — while women who avoid AI entirely may fall behind in building the skills the next labor market will demand.
The World Economic Forum's Global Gender Gap Report 2025 estimated it will take 123 more years to reach gender parity by current trends. AI-driven displacement in clerical and administrative occupations threatens to make that timeline longer.
For insurance industry leaders, the Brookings research functions as a targeting tool — a way to identify which parts of the workforce are most exposed to dislocation and least equipped to manage it, before that dislocation arrives.
Several implications stand out.
Reskilling cannot wait. The occupations at greatest risk are not low performers - they are often the most reliable and longest-tenured members of an organization. Office clerks, claims processors, and administrative assistants who have spent careers building institutional knowledge represent a resource that cannot simply be shed and replaced. Carriers that invest in transition programs now - retraining experienced staff for compliance, analytics, or client-facing roles - will recover that investment in retention, institutional continuity, and morale.
Entry-level pipelines are narrowing. Many of the roles most exposed to AI automation have historically served as entry points into the insurance industry - the first jobs that gave younger workers exposure to claims, underwriting, and customer service before they moved up. As those roles contract, the industry's talent pipeline for experienced professionals could hollow out over time. This is a strategic risk as much as a human one.
The geography matters. The Brookings data shows that vulnerability is concentrated in smaller markets — exactly the places where independent agencies and regional carriers are most prevalent. For those operators, the human cost of AI-driven displacement is not an abstraction. It is their workforce, their community, and their customer base.
A new product category is emerging. The industry should also note that AI-driven job displacement is beginning to create new insurance demand. At least one firm, Singularity, has introduced a parametric AI job loss product - SingularityShield Income Cover - that pays out when an AI-Displacement Risk Index threshold is crossed and a separation notice is filed, delivering up to 50% of net pay for up to 12 months. Whether this product category grows will depend on how quickly displacement accelerates, but the emergence of AI displacement insurance as a line is itself a signal of the moment the industry is in.
There is an irony at the heart of the Brookings analysis worth sitting with. The workers most often cited in public discourse as being at risk from AI - software developers, lawyers, financial analysts - tend to have exactly the savings, skill breadth, professional networks, and geographic flexibility to absorb a job transition. The workers who will struggle most are those who have rarely been the focus of the public conversation: the receptionists, the claims clerks, the policy processors, the administrative assistants - the people, mostly women, who have been running the back offices of American enterprise for generations.
The insurance industry, more than most sectors, has the actuarial tools to model probability and severity of loss. It understands better than almost anyone that the risks that are hardest to see coming are often the ones that matter most.
The data, now, is visible. The question is what the industry does next.
