5 questions about the insurance market

It’s an orderly softening cycle, with more discipline in the marketplace.

There’s an abundance of capital in the marketplace, and top-tier insurers are in favorable reserve positions. Reinsurance pricing is helping, particularly in property. The casualty market is paying close attention to long-term loss cost trends. At the same time, policyholders are operating in a more complex environment, with geopolitical volatility, labor shortages, supply chain disruption, rising material costs, and AI highlighting how interconnected risk has become.

This isn’t a one-market story. We’re seeing greater segmentation and selectivity. Underwriting is shifting away from broad approaches and toward more granular divisions of the market, rooted in the interdependent nature of risk.

Emerging risks are also challenging traditional models. In many areas, the industry does not yet have decades of loss experience to rely on, but there are still opportunities in the market for companies that demonstrate strong risk management.

When policyholders hear “discipline,” they may assume it means retrenchment or saying no. However, it’s less about saying no and more about refusing to accept the unknown.

Historically, carriers may have given policyholders the benefit of the doubt based on class of business. Today, particularly with emerging risks, there is less historical data to draw from and more uncertainty to evaluate. Discipline means deconstructing that uncertainty and understanding its implications for operational risk.

There is often a hidden layer of risk that doesn’t appear in historical loss data; AI is a good example. In this environment, we’re focused on generating sustainable returns and continuing to rely on underwriting profit to support long-term durability.

Discipline is not only about pricing. Terms, conditions, and program structure are critical, especially when pricing signals are less clear. Contract certainty is highly relevant for both underwriters and policyholders.

It’s a factor, but it’s not driving the market.

Risk selection requires realistic assumptions about loss trends not only at a macro level but also at a granular level. We need to understand how trends are playing out across industries and microsectors.

Take supply chain disruption as an example. It’s often thought of as a manufacturing or construction issue, but in reality it cuts across multiple sectors, each with different levels of economic sensitivity. When viewed in isolation, those exposures may seem manageable; however, when they are interconnected across the broader value chain, they can compound and create disruption at a much larger scale.

That’s why uncertainty is rarely uniform. It doesn’t affect every industry or line the same way. Rather than focusing broadly on trends such as inflation, we focus on why certain sectors are more exposed. At the same time, some parts of the economy, like digital infrastructure and life sciences, continue to show durability.

That diversity is a strength of our industry. Our role is to bring insight that helps policyholders understand their environment and build resilience.

There’s a tendency to focus on AI’s efficiencies without fully appreciating its risks.

AI introduces a class of liability for which the industry doesn’t yet have credible loss history. Some companies are deploying AI without mature governance, creating operational, professional, and reputational exposures.

In that sense, AI governance is becoming the new risk control standard. One of the central questions: Who owns the liability when AI gets it wrong — the developer, the user, or another party? Addressing that requires a clear strategy, including defined use cases, oversight of data and users, accountability for outputs, and clarity around decision ownership.

From an insurance standpoint, coverage intent and policy language may start to diverge. Legacy contracts weren’t designed for nondeterministic AI outputs, which reinforces the need for disciplined underwriting and clear terms.

As for the second part of the question, AI is transforming insurance, but it only makes sense if you apply it with discipline and clarity. At CNA, we don’t use AI to replace expertise, we use it to amplify our expertise. It should enhance accuracy and elevate judgment, but it’s certainly not to undermine or substitute it.

For underwriters, the opportunity is to spend more time on high-value underwriting analysis and decision-making. AI can help gather information in a more efficient way with less manual work, enabling high-value underwriting decisions rather than replacing judgment.

The winners in this industry are going to be the ones who blend specialized talent with specialized technology. Moving forward, specialization will be paramount, especially at the industry level.