Climate change is putting insurance models under growing pressure as rising claims costs, natural catastrophe losses and more expensive repairs push premiums higher. For insurers, the challenge is increasingly about more than pricing risk accurately. They also need to find ways to keep cover affordable while maintaining the financial sustainability of their portfolios.
An analysis from Earnix highlights how technology is becoming part of that response, with 95% of French insurers reporting that they have adopted artificial intelligence (AI), while only 32% say they trust their underlying data. The figures point to a growing gap between insurers’ use of AI and their confidence in the information supporting those systems.
France provides a clear example of the pressure facing the industry. Home insurance premiums increased by close to 10% in 2025, while the country’s Cat Nat surcharge increased from 12% to 20% to help replenish reserves within its national natural catastrophe scheme.
The costs behind those increases are significant. Between 1982 and 2024, cumulative Cat Nat claims in France reached €61.2bn. Research from CCR, based on Meteo-France data, estimates that climate change could increase natural catastrophe costs by 40% by 2050. When higher asset values and greater territorial exposure are included, the increase could reach 60%.
The impact is already being felt through extreme weather events. Wildfires in France in 2026 burned almost 100,000 hectares and generated more than €500m in losses, according to figures cited in the source analysis.
As these costs increase, insurers face a difficult balance between accurately reflecting risk through premiums and ensuring customers can continue to afford protection.
French Institute of Actuaries president Simon Le Dily has argued that insurance can remain technically viable while becoming economically or socially unworkable if premiums, deductibles or coverage gaps reach levels that policyholders are unwilling or unable to accept.
There are already signs that customers are becoming more focused on how insurers determine what they pay. A survey of 368 insurance professionals by Transformers de la relation client found that 57.2% identified transparency around pricing and underwriting decisions as the most important factor in customer relationships. This was followed by alignment between price and compensation at 43.2% and price competitiveness at 42%.
For insurers, this means higher premiums need to be accompanied by greater clarity. As the cost of covering certain risks increases, customers need to understand what is driving those changes and what they are receiving in exchange for the additional cost. Technology could have a role to play in addressing both sides of the challenge.
According to Earnix, 33% of insurance executives are already using AI to simulate risk scenarios, while 38% are using the technology to personalise customer recommendations. For climate-related risks, this could allow insurers to model increasingly complex scenarios, identify changing patterns of exposure and develop more targeted approaches to prevention.
However, introducing AI does not automatically solve the underlying data challenge. If insurers do not have confidence in the data being used, more advanced models could simply make existing problems harder to identify.
Data quality and governance are therefore becoming increasingly important as insurers seek to use technology to respond to climate risk.
The issue is particularly significant because the pressure on insurance markets is not expected to disappear. ACPR stress tests indicate that the French insurance sector can remain solvent through to 2050, but this assumes substantial increases in premiums.
That raises concerns about the longer-term accessibility of cover. Higher premiums could lead some customers to reduce their protection or leave the insurance market altogether, potentially shrinking risk pools and increasing costs for those who remain.
At the same time, insurers may become increasingly reluctant to cover areas where climate exposure is too concentrated or difficult to model. France’s insurance system partly addresses this through the relationship between private insurers and state-backed support. CCR chief executive Édouard Vieillefond has described CCR’s role as a systemic shock absorber when risks reach the limits of private insurability.
This approach has helped maintain wider availability of insurance than in some markets where insurers have withdrawn from high-risk areas or introduced broader coverage exclusions. But technology could increasingly determine how effectively the industry can manage the risks that remain commercially viable.
More detailed risk mapping, better use of customer and environmental data and AI-supported prevention could help insurers move beyond simply increasing premiums as losses rise.
Instead, the technology could allow insurers to identify where risks are changing, understand which policyholders are most exposed and encourage preventative action before an event occurs. That could become increasingly important as the industry attempts to address the affordability problem without undermining the economics of insurance.
The challenge is ultimately not just whether insurers can build more sophisticated models. It is whether those models can produce decisions that are accurate enough to support sustainable pricing while remaining understandable and trusted by customers.
Matmut Group director of information systems transformation François-Xavier Enderle said, “Customer data sovereignty is the backbone of the insurance model.”
As climate change continues to increase the cost and complexity of insurance risk, the ability to use data responsibly could become just as important as the AI technology itself.
For insurers, the long-term question is therefore how to combine better risk modelling and prevention with pricing that reflects the changing environment without making protection inaccessible to the customers who need it most. Earnix’s analysis points to AI, scenario modelling and personalised recommendations as some of the tools already being used, but their effectiveness will ultimately depend on the quality of the data behind them and the trust of the customers they are designed to serve.
Copyright © 2026 FinTech Global
