Insurance models are being tested by a risk environment in which exposures increasingly overlap, evolve quickly and create consequences beyond their original category. Earnix, an InsurTech provider, argues that this is putting pressure on the historical data and actuarial approaches insurers have traditionally used to assess and price risk.
The challenge is not simply the emergence of more risks. Cyberattacks, climate change, technological dependency, geopolitical instability, artificial intelligence, data risk and cloud infrastructure vulnerabilities can interact, creating knock-on effects that are difficult to assess independently. Some of these exposures also lack enough historical data to fit conventional actuarial models, making it harder for insurers to determine how and where they can be underwritten.
Cyber risk provides one example of how exposure can spread across multiple areas. An incident can move beyond an initial technology failure to create operational disruption, regulatory consequences, financial losses and reputational damage. The potential impact can increase further where critical infrastructure or interconnected supply chains are affected.
Climate risk presents a similar modelling challenge. Its consequences can extend beyond physical damage into supply chains, migration, social stability and public policy. France Assureurs’ 2026 Forward-Looking Risk Map reflects the changing risk landscape, identifying increasing connections between threats that have traditionally been assessed separately.
These developments are changing the tools insurers use to model risk. Hybrid models, machine learning, stochastic techniques and continuous recalibration are being used to address exposures where historical loss data provides a less complete picture.
Covéa Group chief data officer Arthur Dénouveaux said, “The real challenge is preparing for what we do not yet know how to model.”
The time horizon for some risks is also shifting. Soil subsidence, heatwaves and flooding have increasingly become immediate operational considerations rather than solely longer-term concerns. CCR, using data from Météo-France, projects natural catastrophe losses could increase by 40% by 2050, rising to as much as 60% when higher asset values and territorial exposure are included.
The implications extend into pricing. A survey of 368 French insurance professionals found that 57.2% considered transparency the most important factor in pricing and underwriting decisions. A further 43.2% identified the relationship between price and coverage, while 42% cited price competitiveness.
Data quality presents another constraint as insurers introduce more AI and analytical tools into underwriting. Earnix’s Insurance Industry Trends Report, based on responses from nearly 400 industry professionals, found that 83% of executives were concerned their AI models were being trained on incomplete or inaccurate data. The report also found that 66% believed poor data quality was slowing decision-making.
For insurers, this creates a problem that sits beyond individual modelling techniques. The ability to assess interconnected risks depends on the quality of the data feeding those models and the speed at which insurers can incorporate changing information into pricing and underwriting decisions.
Matmut Group director of IT transformation François-Xavier Enderlé said, “A world without insurance is a profoundly unequal world.”
Earnix’s analysis ultimately points to a broader shift in insurance risk management: emerging exposures are testing not just individual actuarial models, but the data, analytics and decision-making infrastructure supporting underwriting as a whole.
Read the full Earnix analysis.
Copyright © 2026 InsurTech Analyst
