For most of its history, insurance has worked on a fairly simple sequence. We assess risk, price it, transfer it and, if something goes wrong, pay the claim. Artificial intelligence is beginning to disturb that sequence.
In June 2026, Aviva revealed that it had identified more than 18,400 suspected fraudulent claims worth £233 million across its brands during the previous year. Increasingly, some of those claims were being supported by AI-generated images and manipulated documents, including fabricated accident scenes and vehicle damage. The insurer, in turn, is using advanced analytics and AI-enabled tools to detect suspicious claims earlier.
It is a neat illustration of the moment insurance finds itself in. AI is making reality easier to fake at precisely the same time that it is giving insurers more sophisticated ways to understand reality. But fraud detection is only one small part of the story.
Across underwriting, claims, distribution, customer service and risk modelling, AI is giving insurers the ability to know more, sooner. Add satellite imagery, telematics, connected devices and real-time data, and a bigger possibility starts to emerge. Insurance may gradually move from being primarily a business that compensates us after something goes wrong to one that helps stop the loss occurring in the first place.
That would be a much more significant change than simply making insurance faster or cheaper. It changes where the industry creates value.
Insurance has a relevance problem
The timing matters because the risks society needs to insure are becoming harder, not easier, to understand.
Aon estimates that natural disasters caused US$260 billion in global economic losses during 2025. Insurance covered US$127 billion, leaving a US$133 billion protection gap. In other words, even in a year when the global protection gap fell to 51%, roughly half of catastrophe losses still went uninsured.
Cyber risk exposes a similar problem. A June 2026 paper from the Bank for International Settlements – BIS’s Financial Stability Institute describes a cyber insurance market struggling with coverage ambiguity, pricing and accumulation risk as digital systems become more interconnected and AI increases the speed and sophistication of threats. McKinsey & Company estimates that less than 1% of global cyber costs are currently insured.
This is an uncomfortable backdrop for an industry whose purpose is to make uncertainty economically manageable. Risk is expanding into areas where historical data is thinner, losses can be highly correlated and yesterday’s assumptions may be a poor guide to tomorrow.
McKinsey’s July 2026 analysis of insurance economics makes the problem particularly clear. Global insurance revenues have grown more slowly than global GDP over the past decade, while expense ratios in property and casualty insurance have remained stubbornly high. The industry has digitised extensively, yet its basic economic architecture has proved remarkably resistant to change.
AI matters because it potentially attacks both sides of that problem. It can lower the cost of running an insurer, but it can also improve the industry’s ability to understand risks that have previously been too uncertain, volatile or expensive to cover confidently.
The bigger opportunity is moving upstream
Most discussion about AI in insurance begins with efficiency. That is understandable. Underwriters can summarise documents faster. Claims teams can triage cases more quickly. Service centres can automate routine enquiries. Fraud teams can identify suspicious patterns across huge datasets.
But the more consequential opportunity sits further upstream.
McKinsey describes this as a move from “pure risk transfer to risk partnership”. Instead of waiting for a loss and then funding the recovery, insurers can increasingly use continuous data to help customers understand, monitor and reduce their exposure before the claim exists.
We can already see the building blocks. Telematics can identify risky driving behaviour and coach drivers while they are on the road. Satellite imagery and weather data can update commercial property risk as conditions change. Connected sensors can detect water leaks, equipment problems or unsafe conditions before they become expensive failures. Health and life insurers can combine data with coaching and early intervention rather than interacting with customers only when a policy is purchased or a claim is lodged.
Data4Risk offers a useful glimpse of where property insurance is heading. Its Data4Home platform combines AI, geospatial information and data such as property values, crime rates and historical claims to assess risks at individual-property level. Its weather tools use meteorological and location data to identify threats including hail, floods, storms, snow and wind.
None of this means insurers suddenly become omniscient. Models can be wrong, data can be incomplete and customers may quite reasonably object to an insurer knowing too much about how they live, drive or run their business. Prevention creates a data bargain: better protection often requires greater visibility.
Even so, the direction is significant. The traditional insurance promise has effectively been: if something bad happens, we will help you recover. Increasingly, there is an opportunity to add another promise: we will help you see the risk coming.
Better prediction could expand what can be insured
There is another consequence that may matter even more.
Insurance pricing contains a price for uncertainty. When an insurer cannot estimate the likelihood or severity of a loss with enough confidence, the rational response is to charge more, restrict the cover or decline the risk altogether.
That is part of the challenge in climate-exposed property, cyber insurance and emerging AI liabilities. The problem is not necessarily that nobody wants the cover. It is that insurers cannot always price the risk with enough confidence to offer it economically.
Better real-time data and more adaptive modelling could begin to change that. If an insurer can ingest new signals continuously, detect changing conditions and improve the accuracy of claims reserving and settlement decisions, it can reduce some of the uncertainty built into the price.
This is where the AI story becomes more interesting than automation. A technology that improves underwriting accuracy may allow an insurer to write risks it previously avoided or make coverage affordable to customers who were previously priced out.
There are obvious limits. Systemic cyber events, shared cloud infrastructure and dependencies on common AI models can create losses that occur across thousands of organisations simultaneously. Better prediction does not magically make correlated risk diversifiable. AI itself is also generating new categories of liability, digital disruption and fraud.
Still, if AI can help insurers distinguish between risks that are genuinely uninsurable and risks that were merely difficult to understand, the addressable market for insurance changes.
Agentic AI may finally get at insurance’s old plumbing
There is a less glamorous obstacle to all of this: much of the insurance industry still runs on technology built for a different era.
Legacy systems are not simply old computers waiting to be replaced. They often contain decades of product rules, pricing logic, exceptions and institutional knowledge that nobody has documented completely. Replacing them can be expensive, slow and risky precisely because organisations do not always know everything the old system is doing until they try to turn it off. Agentic AI could make that work materially easier.
In an April 2026 analysis, McKinsey described AI agents being used across insurance core-system modernisation to interpret legacy artefacts, reverse-engineer policy logic, generate mappings and code, run tests, reconcile results and coordinate workflows. Across different stages, McKinsey reports typical productivity improvements in its experience ranging from 10% to 90%, depending on the task and degree of automation.
Those figures should not be mistaken for an industry-wide guarantee. But the underlying capability is important. Work that once required scarce specialists to spend months deciphering old code can increasingly be accelerated by systems capable of converting embedded logic into explainable documentation.
This is the difference between putting AI on top of an old process and allowing AI to help redesign the process itself. Previous waves of technology often automated individual tasks while leaving the surrounding operating model intact. Agentic systems have the potential to work across the hand-offs between tasks.
For insurers, that could matter enormously. There is little point having brilliant predictive models at the front end if they remain connected to brittle systems that take months to change.
The customer interface is up for grabs
The transformation does not stop inside the insurer. AI could also change how customers find, compare and buy insurance.
Insurance remains heavily intermediated. McKinsey estimates that around 85% of US property and casualty premiums and 95% of life premiums are still distributed through agents, brokers and managing general agents. Decades of digitisation have changed the tools, but not eliminated the importance of advice and distribution.
Agentic AI introduces a different kind of intermediary. An AI assistant can potentially monitor renewal dates, compare prices and coverage, query carrier systems and recommend a switch before the customer has even thought about shopping around.
This is no longer entirely hypothetical. In 2026, Aviva launched an app in ChatGPT that allows customers to obtain an initial home insurance quote and subsequently expanded it to life insurance quotes. The transaction still moves into Aviva’s own systems to complete the purchase, but the front door has already shifted.
For simpler products, this could weaken one of insurance’s longstanding advantages: inertia. If a customer’s AI agent can continuously scan the market, the effort involved in comparing and switching falls dramatically.
For brokers and advisers, however, the outlook is more nuanced. Complex commercial, specialty and high-net-worth risks still involve judgement, bespoke structuring and advocacy when a claim becomes difficult. AI can make the broker more productive without making the broker irrelevant.
The source of value simply moves. Access to information becomes less scarce. The ability to interpret complexity, challenge assumptions, structure unusual risks and advocate for a client becomes easier to see.
Trust will become more granular
Insurance has always sold something unusually intangible: a promise that will be tested at some unknown point in the future, often when the customer is under stress.
That makes trust central to the business, but AI may force insurers to become more precise about what kind of trust they are trying to earn.
McKinsey’s 2026 insurance analysis distinguishes between trust in the human relationship, trust in institutional credentials and trust in explanation and advice. AI will affect each differently.
There will be moments when human connection becomes more valuable precisely because so much else is automated. A family navigating a life insurance claim, a business recovering from a catastrophic loss or a customer disputing a complex decision may not want an efficient interface. They may want a capable person who understands what is at stake.
Yet AI may outperform today’s service model in other forms of trust. A well-designed assistant can be available at midnight, patiently explain an exclusion three different ways and answer a question a customer might feel embarrassed asking a human adviser. It can make an opaque product easier to understand.
That combination matters because insurance has historically struggled with customer centricity. In a 2025 discussion with leaders from AIA and the Global Asia Insurance Partnership, McKinsey senior partner Bernhard Kotanko argued that insurers need to move away from product-pushing and towards more transparent, personalised solutions and a more direct relationship with customers.
AI can help with that, but personalisation without permission can quickly feel like surveillance. The winners will not simply be the insurers that know the most about their customers. They will be the ones that make clear why they need the information, how it benefits the customer and where human judgement remains accountable.
The future of insurance begins before the claim
There is a useful way to think about the shift underway: Predict. Prevent. Protect. Recover.
Insurance has traditionally concentrated much of its value in the final two stages. It prices and transfers risk, then provides capital and support when the loss occurs. AI allows much more of the industry’s activity to move into the first two.
That does not make claims less important. The moment of loss will remain the point at which an insurer’s promise is most brutally tested. Nor does it mean every insurer will become a risk-management company or every policy will be continuously personalised. Regulation, privacy, economics and customer preferences will put sensible limits around what is possible.
But I suspect the direction is difficult to ignore. The more accurately insurers can detect changing risk, the harder it becomes to justify waiting passively for that risk to turn into a loss.
The most valuable insurer of the next decade may therefore not be the one that processes the claim fastest. It may be the one that helps ensure fewer claims need to be made in the first place.
For an industry built around being there when something goes wrong, that is a significant change in where the relationship begins.
Michael McQueen is a globally recognised trend forecaster, change strategist and keynote speaker.
A bestselling author of 10 books, Michael’s latest release was named by Malcolm Gladwell and Adam Grant as one of the top five must-read new leadership books. He is a sought-after media commentator, with his insights regularly featured in Forbes, The Guardian, and CNN.
To find out more about Michael and his work, click here.
NOTES
Aon. “2026 Climate and Catastrophe Insight.” Aon, 2026. https://www.aon.com/en/insights/reports/climate-and-catastrophe-report.
Aviva plc. “Aviva Stops Record Levels of Claims Fraud as Scams Become More Sophisticated.” June 8, 2026. https://www.aviva.com/newsroom/news-and-research-overview/news-releases/2026/06/aviva-stops-record-levels-of-claims-fraud-as-scams-become-more-sophisticated/.
Aviva plc. “Aviva Launches Insurance App on ChatGPT.” March 24, 2026. https://www.aviva.com/newsroom/news-and-research-overview/news-releases/2026/03/aviva-launches-insurance-app-on-chatgpt/.
Aviva plc. “Aviva Expands ChatGPT App to Life Insurance Applications.” June 1, 2026. https://www.aviva.com/newsroom/news-and-research-overview/news-releases/2026/06/aviva-expands-chatgpt-app-to-life-insurance-applications/.
Currat, Adrien, Joe Perry, and Jeffery Yong. “Cyber Insurance Unpacked: The Corporate Digital Safety Net.” FSI Insights no. 75. Basel: Financial Stability Institute, Bank for International Settlements, June 2026. https://www.bis.org/publications/fsi-insight-75-cyber-insurance-unpacked-corporate-digital-safety-net.
Data4Risk. “Data4Home: Data Solutions for Insurance Risk Management.” Accessed September 25, 2026. https://www.data4risk.com/data-4-home/.
McKinsey & Company. “Can Agentic AI (Finally) Modernize Core Technologies in Insurance?” April 29, 2026. https://www.mckinsey.com/middle-east/our-insights/can-agentic-ai-finally-modernize-core-technologies-in-insurance.
Ralph, Jason, Johannes-Tobias Lorenz, Nick Milinkovich, Sid Kamath, Tanguy Catlin, and Gabriella Meijer. “How AI Will Reshape the Economics of Insurance: A CEO’s Guide to Strategy.” McKinsey & Company, July 23, 2026. https://www.mckinsey.com/industries/financial-services/our-insights/how-ai-will-reshape-the-economics-of-insurance-a-ceos-guide-to-strategy.
Ye, Jady. “The Future of Insurance Is Personal: Insights from Asia’s Industry Leaders.” Interview with Min Hung Cheng, Stuart Spencer, and Bernhard Kotanko. McKinsey & Company, October 21, 2025. https://www.mckinsey.com/industries/financial-services/our-insights/the-future-of-insurance-is-personal-insights-from-asias-industry-leaders.
