Since 2019, several casualty lines of business have shown a consistent pattern of adverse development through year-end 2024. Preliminary third-quarter 2025 disclosures signal that this trend is not yet reversing. The underlying experience, however, differs significantly by line and by carrier. This article focuses on where and why reserve shortfalls are occurring. We also provide some high-level suggestions to adjust actuarial methods for more adequate reserves.

Using Annual Statement data from S&P Global Market Intelligence, we examined:

• Commercial auto liability (industrywide), and
• Other liability—occurrence experience for 20 writers concentrating on excess or umbrella coverage.

Across accident years 2016–2024, published ultimate loss ratios have increased almost every calendar year. With the benefit of hindsight, initial and subsequent reserves were inadequate.

From Annual Statement data via S&P Global Market Intelligence Industry Commercial Auto Liability

From Annual Statement data via S&P Global Market Intelligence based on Other Liability Occurrence results from 20 companies that predominately write excess and umbrella business.

What is driving this pattern of inadequate reserves? We believe that the following factors are the most significant:

1) Extended Litigation: The expansion of third-party litigation funding and the improved capitalization of certain plaintiff firms mean more lawsuits proceed to trial. This causes challenges with traditional actuarial methods. Actuaries often use the past patterns to predict future patterns; however, if the environment changes significantly the methods become less reliable. With an increasing percentage of claims being litigated, historical loss emergence patterns are less reliably predictive of the future patterns. The industry has observed both longer cycle times (from claim report to claim settlement) due to more litigation and increased settlement costs as jury outcomes increasingly favor plaintiffs.

2) Backward-looking benchmarks: Actuaries often use older years' loss ratios to estimate loss ratio results for more recent years (after adjusting for premium changes and loss trends). However, if the older years' loss ratios consistently increase, the initial assumptions for the newer years start too low.

3) Under-estimated trend in a rising-cost environment: In an environment of increasing costs, it is difficult to estimate trend factors. For example, if average claim costs are increasing, some companies may believe that case reserves are more adequate and therefore not reflect the higher trends in the projections.

4) Management optimism. After the large rate increases and underwriting tightening during 2019-2022, some management teams find it hard to believe that loss ratios are not dramatically improving. This belief can delay the recognition of continuing adverse development.

The published industry results for the last few years clearly indicated adverse industry development as illustrated in the graphs above. Preliminary data published through the third quarter of 2025 indicates adverse development is continuing for some companies.

The table below displays development through the third quarter for all lines of business, separated by companies that indicated favorable development for accident years 2022 and prior and those that indicated adverse development.

*Based on Accident Years 2022 and Prior

From Annual and Quarterly Statement data via S&P Global Market Intelligence

For the companies we have summarized that reported third-quarter data, this industry composite displayed little change in prior year reserves for accident years 2022 and prior, with favorable reserve development for accident years 2023 and 2024. 53% of the companies indicated favorable development and 47% of the companies indicated adverse development for accident years 2022 and prior. We note that reserve development differs by company in the amount and magnitude due to the lines of business written.

The quarterly data reported to the NAIC is not presented in the same level of detail as the year-end data, as Quarterly Statements display development for all lines of business combined. Therefore, we segregated the companies into different groupings based on our assessment of the type of business the companies write. Additionally, the quarterly development is only available for accident years 2022 and prior, 2023 and 2024.

The cohorts of companies that primarily write personal lines business, workers compensation business, medical malpractice business and mortgage insurance displayed favorable reserve development for accident years 2022 and prior, and also for accident years 2023 and 2024. Personal lines business as well as workers compensation business are lines generally less affected by social inflation. For accident years 2022 and prior, the total combined reserves for these cohorts of companies developed favorably by approximately 3%.

From Annual and Quarterly Statement data via S&P Global Market Intelligence

However, the cohort of companies that write primarily commercial insurance, companies in run-off, and reinsurance companies displayed adverse development for accident years 2022 and prior.

From Annual Statement data via S&P Global Market Intelligence

Drilling down within the commercial lines writers provides additional insights. The following table displays the reserve development for commercial lines writers that:

• write limited amounts of workers compensation;
• write both commercial and personal lines;
• are excess and surplus lines companies; and
• are writers of other commercial lines of business including workers compensation (i.e., "other commercial writers").

The cohort of companies that primarily write commercial lines with limited workers compensation business displayed higher adverse development (2.6% of adverse development for accident years 2022 and prior) compared to their more diversified peers that also wrote either workers compensation or personal lines (these cohorts displayed 0.3% of adverse development for accident years 2022 and prior). It is reasonable to assume that commercial lines carriers that are more diversified (e.g., write workers compensation or personal lines business) are benefiting from the favorable development on these lines which mitigates the development they may be experiencing in their commercial business.

The cohort of commercial companies with limited workers compensation business also write commercial automobile liability. The companies that primarily write commercial auto liability are displaying higher adverse development. We did not separately segregate these companies as the reserve base is limited and the development is driven by a few companies. Commercial auto liability is a line of business more affected by social inflation that has had significant rate increases and re-underwriting over the past few years, which increases the uncertainty in the reserve estimation process.

We note there is variability within the various cohorts and for certain cohorts of companies, a few large carriers had a significant effect. Within the "favorable" cohorts, 41% of companies posted adverse development for accident years 2022 and prior. Conversely, 49% of insurers in "adverse" cohorts reported favorable development for those same accident years. For the lines of business affected by social inflation, prior years' development hinges on how effectively each insurer has captured social inflation effects in past analyses and how aggressively they are recognizing those pressures today.

Favorable cohorts: Companies writing personal lines, workers compensation, medical malpractice and mortgage insurance

Adverse cohorts: Companies writing commercial business, companies in run-off and reinsurers

Given the factors outlined, we expect unfavorable reserve development to persist for certain lines of business and companies. However, favorable and adverse development will affect insurance carriers differently depending on the lines of business they write and their prior recognition of social inflation in the actuarial methods.

Although accident years 2023 and 2024 are generally indicating a favorable run-off, we have a concern that adverse development will occur in these accident years as the historical adverse development may not be fully reflected in the actuarial assumptions.

To reflect social inflation in actuarial methods, we recommend companies:

• Reevaluate the expected loss ratios that are used in actuarial methods to not only reflect historical adverse development but also current claim activity; and
• Separate lines of business into more granular groupings which segregate those segments more affected by social inflation and those less affected by social inflation (e.g., litigated versus non-litigated claims).

After year-end 2025 data is released, we will publish a companion article that presents updated results with more details by line of business, along with greater discussion on how to adjust actuarial methods.

Today's 25-year-olds are far less likely to have life insurance than the 25-year-olds from the late 20th century. In 1989, for example, 65% of adults aged 20 to 39 owned some form of life insurance; by 2025, that total dropped to as low as 36% of Gen Z members.

Though part of a general trend toward less life insurance ownership today (Figure 1), the reasons behind the decline among younger consumers are distinct, according to LIMRA and Capgemini's recent World Life Insurance Report 2026. Among them: Historical triggers such as marriage and parenthood aren't happening as soon for today's under-40 consumer – if they happen at all.

This is just the latest data point that spotlights the challenge of selling insurance to the next generation.

For insurers seeking to attract and turn young customers into long-term policyholders, it is essential to develop tailored solutions that address their needs and perceived challenges. RGA's research and experience reveal four key design considerations – relevance, simplicity, affordability, and flexibility. By incorporating these four factors into product development, insurers are starting to find success in attracting younger generations.

Why are young consumers not buying life insurance?

Delays in life milestones that historically prompted insurance purchases present a challenge for the industry. The average age at which a man and a woman marry today has increased approximately eight years since 1960, from 22.8 to 30.5 for men and 20.3 to 28.1 for women.

The LIMRA survey also revealed that 63% of consumers under the age of 40 have no immediate marriage plans, and 84% of both single and married under-40 adults have no immediate plans to have a child. This could be a contributing factor behind data showing Gen Z life insurance ownership is more than 20 percentage points lower than it is for Baby Boomers, and 14 percentage points lower than their own next-closest generation, Millennials (Figure 3).

This decline is not due to lack of intent. More than a third (39%) of consumers said in a 2023 U.S. survey they intend to purchase life insurance in the coming year, with higher numbers for Gen Z (44%). Yet those intentions are not turning into actual purchases in sufficient numbers to halt or reverse declining sales.

The reasons younger consumers give for failing to purchase the coverage they know they need suggest opportunities for insurers to develop new solutions addressing those challenges, which include:

• Concerns around affordability, real or perceived
• Uncertainty about protection needs and what products to purchase, indicating potential product complexity or lack of relevance from existing offerings
• Lack of access, pointing to the limitations of traditional life insurance distribution models for reaching that demographic, as evidenced by the larger share of Gen Z and Millennials saying that no one has approached them about life insurance, or that it is not offered by their employer, relative to survey respondents from older generations

Designing life insurance solutions for younger consumers

Insurers have an opportunity to design life insurance propositions that are tailored to the needs of younger customers and that address the concerns they have identified. In doing so, the following design principles should be considered:

• Relevance of the offering, including the core insurance benefits, value-added services or perks, and the distribution approach
• Simplicity, including easy-to-understand insurance benefits and terms, authentic language, and streamlined processes
• Affordability of the overall proposition, both real and perceived
• Flexibility of the coverage and payment options

Relevance

Young consumers are not a monolithic group. They range in stage of life from those just graduating high school through those who could be considered mid-career professionals.

Insurers can design and offer more relevant insurance products that align with key milestones and associated protection needs across this timeline, including life insurance products that offer value-added services relevant to younger consumers. Distribution approaches may also need to be adapted to different life stages, to help mitigate the concerns mentioned by younger customers about lack of access. This may involve considering alternative distribution approaches through partnerships with companies that already have a relationship with younger customers, rather than relying exclusively on traditional distribution models.

For example, home prices in India have increased 1,500% in the past three decades, according to one estimate. That has led more people to delay home purchases and rent instead. One digital real estate marketplace in India partnered with an embedded insurance startup to offer a rent protection plan that can be purchased with a click of a button when consumers submit their online rental payments. Premiums are embedded into the monthly rent payment workflow. Benefits include critical illness coverage from 15 conditions, a personal accident payment, and medical expenses in case of accidental hospitalization.

Increased relevance can lead to growth in the younger-consumer market. Once this younger generation is familiar with a relevant, affordable form of coverage, they can grow into other products as new life events occur and protection needs emerge – from home purchases to marriage to children.

Simplicity

Young consumers have indicated their uncertainty about how much coverage they need and which product to buy (Figure 4). This is likely in part due to the perceived complexity of existing insurance products, and of the language used to describe policy benefits and conditions. Insurers have an opportunity to design simple and easy-to-understand products that young consumers may be more comfortable purchasing, particularly in cases where they have limited access to advisors.

There is also an opportunity to leverage behavioral science techniques to increase comprehension. RGA's research shows that making key policy details more salient, such as by using FAQs or visuals, and improving the relevance of the information presented, for example by leveraging tools and calculators, as well as using video content, can significantly improve comprehension and engagement in digital life insurance customer journeys. These strategies should be particularly relevant for younger demographics accustomed to consuming visual content on social media platforms.

Furthermore, real-world case studies show that improved comprehension through behavioral science techniques can lead to a 48% increase in policy renewal rates and a 32% reduction in policy cancellation rates.

Insurers also need to continue streamlining the life insurance purchase journey they offer to match younger consumers' expectations based on the experiences they're accustomed to from the apps they use daily. As an example, a digital life insurance distributor partnered with RGA and insurers in Canada to offer easy-to-purchase term life and critical illness insurance products, which have achieved success in the market, particularly with younger customers. They offer a streamlined digital customer journey, which allows users to complete a jargon-free fully underwritten application in as little as 20 minutes and receive an instant decision if eligible.

Affordability

The insurance industry is battling a perception problem. One of the most frequently cited reasons for not having life insurance is that it is too expensive. But when asked for a cost estimate, younger consumers are far more likely to give a price far above the median than older consumers.

Innovations attached to low-cost products offer a direct way to change the narrative.

For example, an RGA-supported effort with a bancassurance product uses banking data to assess the risk of applicants and offer discounts to those who prove to be better risks. This tends to favor younger consumers, who generally are also healthier. Discounts range up to approximately 20% for the lowest risk segments.

Insurers can also address affordability concerns by offering benefits that young consumers appreciate and are able to use today, enhancing the overall value proposition of life insurance. For example, a life product sold through one of the largest retailers in Spain incorporates a wellness rewards program that provides incentives for physical activity. The product includes a mobile app that tracks daily steps and syncs with popular activity trackers. Clients can earn maximum rewards by averaging 10,000 or more steps per day. The program rewards users for maintaining an active lifestyle by converting their daily steps into monetary rewards, which can be accumulated and redeemed as gift cards.

The offering is tied to a popular shopping destination, and is the first insurance product in Spain rewarding customers for wellness activity. It attracts younger people by offering a relevant, simple, and affordable living benefit that encourages healthier living through flexible rewards.

Some items insurers must consider in balancing costs and customer cover include:

• Full underwriting vs. simplified (or even guaranteed) issue
• Level vs. YRT premiums, including any premium guarantees
• Basic life cover vs. accidental cover vs. comprehensive covers, including critical illness and disability

This leads directly to the final consideration.

Flexibility

People tend to procrastinate until they see a solution that matches their unique needs. Young people need to see insurance products relevant to them. Each customer has specific needs that must be balanced in coverage type and levels, premium structures, and underwriting.

For example, offering low and affordable covers initially that can flex as life events occur – such as home purchases, marriages, parenthood – or have the option to renew or convert into further insurance covers can be seen by younger consumers as relevant to their lives right now and, thus, individually tailored.

Similarly, while full medical underwriting traditionally leads to the lowest premium for healthy applicants, it can be too complex, and a barrier to entry, for younger consumers used to quick purchases. Offering flexible options that accommodate different trade-offs between price and customer experience can help address the wide range of needs expressed by younger consumers, from the most cost-sensitive, to those prioritizing the smoothest customer journey.

Conclusion: Gaining customers for life

The insurance industry is seeing a change in life triggers, not an evaporation. Marriage and children might be happening later, but other milestones exist for insurers to offer relevant, simple, affordable, and flexible protection products.

Younger consumers may be harder to convert than in past generations, but product innovations can create a rewarding win-win proposition for both the young policy holder and the insurer.

As of early 2026, artificial intelligence (AI) has made only a modest dent in the daily practice of law. Adoption is rising, but cautiously. Many lawyers still avoid AI altogether; others limit its use to narrow, low-risk applications. This restraint sits uneasily alongside the predictions that have circulated since early 2023, when evidence emerged that ChatGPT could earn passing grades on law school exams—and even on the bar exam.

The gap between promise and practice has fed a familiar narrative: if AI were truly transformative, law firms would already look different. Because they largely do not, skeptics argue that AI's disruptive potential has been oversold.

That conclusion gets the timing wrong — and misunderstands what disruption in law actually looks like.

From the beginning, strong performance on exams was a poor proxy for real-world impact. Legal practice is governed by ethical obligations, professional judgment, and client risk—not multiple-choice questions. Lawyers must closely review AI-generated output, especially given well-documented risks of hallucinations and subtle errors. In many contexts, reviewing and correcting AI-assisted work has been slower than producing it directly, or has resulted in lower-quality outcomes than human-only work—particularly when the lawyers involved are highly skilled or when precision matters more than cost savings.

The problem, in short, was not overhype. It was the wrong benchmark.

The question that actually matters is not whether AI can perform legal tasks on its own, but whether lawyers using AI outperform comparable lawyers who do not. Until recently, the answer to that question was unclear at best. Now, emerging evidence suggests it is beginning to turn decisively in AI's favor.

In a recent study, my colleagues and I reported the first randomized controlled trial evaluating two AI innovations with direct implications for legal work. The first is Retrieval-Augmented Generation (RAG), which grounds AI outputs in authoritative legal sources rather than free-form text. The second is the rise of AI reasoning models that structure complex analysis before generating responses.

In the study, upper-level law students were randomly assigned to complete realistic legal tasks using either a RAG-enabled legal AI system, an AI reasoning model, or no AI assistance at all. The results mark a clear break from earlier studies focused on prior generations of large language models. Across multiple tasks, participants using modern AI tools produced meaningfully higher-quality legal work. They also worked much faster. In five of six tasks, productivity increased by between 50% and 130%.

This evidence reframes the AI debate in law. The story is no longer about machines replacing lawyers—or about AI's ability to ace exams. It is about augmentation that finally works: tools that allow lawyers to do better work, in less time, without sacrificing professional standards.

That shift carries real consequences for legal institutions.

Firms that continue to treat AI as an experimental add-on or a compliance risk may soon find themselves at a competitive disadvantage. If AI-enabled lawyers can reliably produce higher-quality work more efficiently, then billing models, staffing decisions, training pipelines, and even partner expectations will come under pressure. Early adopters will not simply save time; they will set new baselines for quality and speed that others will be forced to match.

The implication is clear. The window for cautious observation is closing. In 2026, the strategic question for law firms is no longer whether AI will meaningfully affect legal practice, but how quickly—and whether they will shape that transition or be shaped by it.

AI chatbots have become a regular part of how insurance services operate. Customers turn to them to review policies, follow claim updates, or get quick answers without sitting on hold. For insurers, they help handle high volumes of requests while keeping support costs in check.

As these tools become more visible, expectations sometimes drift. Not every task in insurance should be automated, and not every interaction benefits from a chatbot. Understanding where AI chatbots should step back is key to using them well.

This article looks at the functions that still require people, even as insurance chatbots and broader AI insurance services continue to evolve.

Where AI chatbots make sense

There's no question that AI chatbots add value in the right places. Tasks that follow clear rules and don't depend on interpretation are a good fit. Simple policy questions, coverage summaries, payment reminders, and claim status updates are all examples where automation works reliably.

That's why many insurers treat chatbots as the first point of contact. They take on routine requests, ease pressure on call centers, and keep basic information accessible at any time. In that role, chatbots help guide users and filter requests, rather than make final decisions.

Issues usually arise when the same tools are pushed into areas that require judgment, risk assessment, or clear accountability.

Why judgment-based decisions still need people

Many insurance decisions live in gray areas. Coverage disputes, claim denials, and policy exceptions are rarely decided by one clear rule. They usually depend on context, intent, and how similar situations were handled before.

A chatbot can help surface the explanation or point to relevant policy language, but it shouldn't be the authority behind the decision. Once financial impact or legal exposure is involved, a person has to be responsible for the outcome. This boundary matters not just for compliance, but for trust.

AI chatbots in the insurance industry and emotional context

Insurance conversations don't always happen at calm moments. Accidents, property damage, and unexpected losses bring stress with them. Customers often need reassurance as much as information.

AI chatbots in the insurance industry can respond politely and quickly, but they don't genuinely understand emotional nuance. They can't sense frustration building or know when a conversation needs to slow down.

In these situations, escalation to a human agent isn't a failure of automation. It's a necessary part of good service design.

Claims handling beyond the simple cases

Some claims are simple and move quickly. Others take time, context, and careful review.

For low-risk cases with clear documentation, chatbots can play a useful role. Once a claim includes conflicting information, unclear responsibility, or a higher financial impact, automation starts to fall short.

At that point, human adjusters are essential. They review evidence, interpret policy wording, and make decisions that may need to be held up later under review or dispute. Chatbots can assist by organizing information, but they shouldn't own the outcome end-to-end.

Why AI fraud detection still needs people

AI fraud detection is often presented as one of the strongest areas for automation, and in many ways it is. Systems can scan large volumes of data and surface unusual patterns far faster than any manual process.

What these systems struggle with is intent. In real-world use, AI works best as an early filter, pointing investigators toward cases that deserve a closer look and leaving the final judgment to people.

Insurance chatbot use cases that need a handoff

Many insurance chatbot use cases work best when they are set up as shared workflows rather than fully automated paths. The chatbot handles the first interaction, gathers the required information, and then routes the case forward.

Policy changes, renewals, endorsements, and compliance-related questions often fall into this group. Rules can vary depending on region, policy type, or specific circumstances, which means final guidance usually needs confirmation from a person.

In these situations, a smooth handoff isn't a limitation. It's what keeps the process accurate, compliant, and trustworthy.

Negotiation is another clear boundary. Settlement discussions, premium adjustments, and special terms require flexibility and judgment that chatbots don't have.

Conclusion. Why knowing the limits matters

It's easy to judge progress by what AI chatbots can handle. In insurance, setting limits often matters more than expanding capabilities. Chatbots work well for simple interactions, but customers expect human involvement when the stakes rise.

Insurers that design around this reality tend to see stronger outcomes from automation. They gain efficiency without giving up control, and they improve service without undermining trust.

A year after the devastating Los Angeles wildfires of January 2025, the insurance and reinsurance industry is still absorbing the scale of their impact, as well as the lessons learned for future risk assessment. What is now clear is that these fires were not an anomaly but a warning signal for how wildfire risk is evolving in a changing climate.

The financial consequences of the LA wildfires were significant. To date, insurers have paid out approximately $22.4 billion, with total insured losses reaching around $40 billion overall. These figures place this firmly among the most costly natural catastrophe events in recent US history, reinforcing wildfire's status as a primary driver of loss rather than a secondary peril.

The sparks behind the blaze

In the year since, investigators have been able to piece together a clearer picture of how the fires began and why they escalated so rapidly. The initial blaze, the Palisades Fire, was started by human activity. Emergency services believed the fire had been successfully extinguished, but a combination of Santa Ana winds and exceptionally dry conditions caused it to re-ignite, with devastating consequences.

The second major event, the Eaton Fire, was ignited by a nearby power line. While these two fires inflicted the most severe damage, they were only two of 14 separate wildfires that occurred across the region during January 2025.

Climate change also played a central role in amplifying their severity. Heavy rainfall during 2022 and 2023 drove extensive vegetation growth across California. This period of abundance was followed by a prolonged drought, which dried out that vegetation and transformed it into highly combustible fuel. In effect, climate volatility, not just warming, created ideal conditions for wildfire spread.

A global shift in wildfire behavior

The human cost of these events has also been profound. While 31 deaths were attributed directly to the fires, a medical study published in JAMA (The Journal of the American Medical Association) estimates that up to 400 additional deaths may have been caused indirectly, driven by poor air quality and reduced access to healthcare during and after the events. Further, more than 100,000 homes were evacuated, disrupting communities and livelihoods on a massive scale.

Taken together, these impacts underscore a sobering reality: wildfire risk is no longer confined to historically defined burn areas or traditional seasonal expectations. The LA fires echoed patterns seen elsewhere, including the Australian bushfires of the same year, where successive wet years followed by extreme heat produced similarly combustible landscapes.

The diversity of ignition sources - human activity, infrastructure failure, weather-driven re-ignition, and climate change - highlights a critical challenge for risk modelling: wildfire cannot be understood through a single causal lens and the parallels seen across hemispheres point to a global shift in wildfire behavior.

What this means for our industry

For underwriters, the key takeaway is clear and urgent: past bushfire and wildfire burn areas are no longer a reliable predictor of future fires. Historical loss data, while still of value, cannot on its own capture the rapidly changing interactions between climate, vegetation, weather extremes, and ignition sources.

These distinctions matter. The latest wildfire models, such as BirdsEyeView's, for instance, explicitly focus on different ignition mechanisms, recognizing that the probability, timing, and severity of fires vary materially depending on how they start and how environmental conditions evolve around them. Treating wildfire as a homogeneous peril obscures these dynamics and increases underwriting blind spots.

This has significant implications for pricing, accumulation management, and capital allocation. Wildfire has firmly shifted from a 'secondary' peril into a core driver of portfolio performance. Models calibrated on assumptions of climate stationarity risk lagging reality at precisely the moment when precision matters most.

Progress lies in prediction

Looking ahead, the industry's ability to adapt will depend on its willingness to embrace climate-aware, data-driven analytics. Advances in satellite observation and machine learning now allow us to monitor fuel load, vegetation stress, and environmental conditions in near real time, enabling earlier detection of emerging risk patterns and more responsive underwriting decisions.

The lessons from the LA wildfires, one year on, are therefore not only about loss – they are about learning. If reinsurers and insurers can move beyond retrospective modelling and adopt adaptive intelligence, they will be far better positioned to navigate the growing volatility of wildfire risk.

In a world where climate dynamics are rewriting the rules, resilience will belong to those who can see risk forming before it ignites.

When the Buffalo Bills fired head coach Sean McDermott on Monday, that brought the number of top jobs vacated this season to 10. That struck me as a huge number, out of just 32 such positions. I'm accustomed to six or seven, maybe even eight. 

It turns out that the 10 departures this year are the most since... the end of the 2022 season. And eight of the 10 hires from three seasons ago have already been fired. 

I realize I have the luxury of surveying the frenzied hiring and firing as a lifelong fan of the Steelers, who have had precisely three head coaches since 1969. (There have been six popes in that same stretch.) The first two of those Steeler coaches are in the Hall of Fame, and Mike Tomlin, who resigned last week after 19 seasons, will surely join them when he becomes eligible. 

It's obviously not helpful to tell teams that they, too, should hire Hall of Fame coaches, but I do think lots of teams are getting one thing very wrong — and anyone leading an innovation effort, including at an insurance company, may be tempted to make the same mistake. 

Basically, too many teams don't have a long-term vision. They think they do, but they don't. So they lose patience too quickly. They get twitchy when the results aren't there immediately and move on to the next coach or general manager or both, only to pull the plug on them too quickly, too. As I wrote two weeks ago — in what I thought would be my one NFL reference for the year — the impatience is partly because owners get focused on the outcomes of their choices, rather than on whether they made a good bet. 

Owners ask: Did we win the Super Bowl this year? They don't ask: Did we put ourselves in a good position to have a chance to win? They don't ask: Are we putting together the pieces for the next several years? They ask: Did we win this year?

That sort of thinking is how you become the Cleveland Browns. Since the 1999 season, when the team was reconstituted after the original franchise moved to Baltimore and became the Ravens, the Browns have had 12 head coaches and are about to hire their 13th. In those 27 seasons, they've won 33% of their games and zero titles in the four-team AFC North. They've played in all of four playoff games, winning one.

Yet the Bills have decided to follow suit. They fired McDermott even though he took the Bills to the playoffs in the last seven seasons and in eight of his nine seasons as head coach. They made it to the conference championship twice, losing both times to the formidable Chiefs. The Bills hadn't been to the playoffs in the 17 years before McDermott arrived. Who do they think they'll get who'll be better? 

Firing Pete Carroll as the head coach of the Las Vegas Raiders after one season? Sure, the Raiders were an awful 3-14 this season, but that's only one game worse than their record for the 2024 season, and they'd had losing records in 2022 and 2023, as well. You're telling me they didn't know what they were getting in a 74-year-old coach who, among many other things, had just coached for 14 seasons in Seattle? I have no idea whether he was the right fit in Las Vegas, but either Raiders ownership was too quick to commit to him as the turnaround guy a year ago or was too quick to bail after this season. In either case, the Raiders showed they're dysfunctional.

By contrast, when the Steelers hired Chuck Noll in 1969, he was a 37-year-old with no track record as a head coach and went 1-13 in his first season. He had losing records in his second and third seasons, too. But he was putting the pieces together, the Rooney family stuck with him, and the magic started happening in season four.

When the Cowboys hired Jimmy Johnson as head coach in 1989, he went 1-15 his first season. But he and owner Jerry Jones had a long-term vision largely based on the eight draft picks, including three first-rounders, they got for trading Herschel Walker to the Vikings. By season four, Dallas was winning the first of the three Super Bowls it took in four years.

The Cowboys actually demonstrate both patience and impatience. After Johnson led the team to two Super Bowls, Jones felt Johnson was getting too much credit. So — in what I believe is the dumbest decision ever by the owner of a sports franchise — Jones fired Johnson and brought in another college coach who had won national championships to try to show that just about anyone could win with the juggernaut Jones, the self-proclaimed genius, had put together. The Cowboys did, in fact, win one more Super Bowl, but then got twitchy as Jones took more control and have never recovered. The Cowboys have won five playoff games in the last 30 seasons and have never even made it back to a conference championship game. 

What Should Insurers Do?

Insurance companies have a luxury that NFL franchises don't: They don't have to deal with hundreds of podcasts by rabid fans who want to fire everybody any time someone fumbles a football. 

Still, insurance companies have to answer to shareholders, and they do have to succeed. That means innovation efforts, especially related to generative AI, need to fit into a long-term vision. They can't be one-offs, because those are too easy to kill. And the efforts can't be judged based just on whether they succeeded or on any other short-term indicators. 

The right questions are: Were they good bets? Did we learn something important? What do we do next to build on what we just learned?

Early on, it was at least okay to do broad experiments with Gen AI. People needed to get comfortable with the concept, and the applicability was somewhat nebulous. But we're more than three years into the Gen AI revolution now, so it's time to do more long-term planning about how Gen AI can both make your organization more efficient and about how it might even let you make more radical changes to your business model. 

Once you've laid out that vision, you have to stick with it. None of this firing the coach or heading off in a new direction the first time something unexpected happens. And the commitment has to be communicated from the top of the organization, repeatedly, so people know this isn't just a phase that they can assume will pass them by if they just keep their heads down. 

I can't guarantee success. Even my Steelers haven't won a playoff game since 2017, and there's no guarantee we won't pick a dud as head coach this time. Dan Rooney played a major role in hiring all three of our coaches since 1969, and he died in 2017. But I can guarantee that taking a stable, long-term approach means you won't be the Cleveland Browns. 

Cheers,

Paul

P.S. How committed have the Steelers been to their head coaches for the long term? My father once told me an illustrative story that was passed on to him by a friend who was the PR guy for the Steelers in the 1970s and 1980s. 

Now, my father was a hail-fellow-well-met, Irish storyteller type, but his stories always started out based on something that actually happened, and I choose to believe this took place just as my father described: 

The PR guy said he was sitting in Noll's office at the end of a workday, when Dan Rooney stuck his nose in. 

Rooney said, "Chuck, I put your contract in your in-box. I left the numbers blank because it's your turn to put them in this year." 

Noll responded, "No, no, it's your turn. I put the numbers in last year."

Rooney said, "I checked. I did the numbers last year. Just put the contract in my in-box when you're done, and I'll sign it in the morning. Have a good night."

Cyber incidents created many headlines in 2025 and are still the biggest worry for companies globally in 2026, according to the just released Allianz Risk Barometer. The past year has also been a significant one for accelerated adoption of artificial intelligence (AI), which is reflected in its ranking as the biggest riser in the annual survey at #2. Close to half of survey respondents believe AI is bringing more benefits to their industry than risks. However, a fifth say the opposite.

The Allianz Risk Barometer is an annual business risk ranking compiled by Allianz Group's corporate insurer Allianz Commercial together with other Allianz entities. Now in its 15th year, the Risk Barometer incorporates the views of 3,338 risk management experts from almost 100 countries and territories and identifies the main perils risk management practitioners are expecting in 2026.

Cyber risks by far the biggest concern for companies

In 2026, cyber incidents are the top global risk for the fifth year in a row, with its highest-ever score (42% of responses), and by a higher margin than ever before (+10%). It ranks as the main corporate concern in every region (Americas, Asia Pacific, Europe, and Africa and Middle East).

The continued presence of cyber at the top of the Allianz Risk Barometer reflects a deepening reliance on digital technology at a time when the cyber threat landscape, and geopolitical and regulatory environments, are fast evolving. Recent high-profile cyber-attacks underline the continuous threat for businesses of all sizes. Smaller and mid-sized enterprises are increasingly targeted and under pressure due to a lack of cyber security resources.

AI creates emerging risks as well as new business opportunities

AI has surged into the top tier of global business concerns, rising to #2 (32%) in 2026 from #10 in 2025 – the biggest jump in this year's ranking. It is a big mover in all regions – ranked #2 in the Americas, Asia Pacific, and Africa and the Middle East, and #3 in Europe – and is a growing risk for companies of all sizes too, moving into the top three for large, mid-sized and smaller firms.

As AI adoption accelerates and becomes more deeply embedded in core business operations, respondents expect AI-related risks to intensify, especially when it comes to liability concerns. The rapid spread of generative and agentic AI systems, paired with their growing real-world use, has raised awareness of just how exposed organizations have become.

Business interruption strongly connected to geopolitical risks

2025 marked a shift towards protectionist trade policies and tariff wars that brought uncertainty to the world economy. It was also a year of regional conflicts in the Middle East and Russia/Ukraine, as well as border disputes between India/Pakistan and Thailand/Cambodia and civil wars in Africa – a trend which continues in 2026 with the U.S. intervention in Venezuela.

Geopolitical risks are putting supply chains under increasing pressure, but as risks rise, just 3% of Allianz Risk Barometer respondents view their supply chains as "very resilient". In the past year alone, trade restrictions have tripled to affect an estimated U.S.$2.7 trillion of merchandise – nearly 20% of global imports according to Allianz Trade – fueling companies exploring trends such as friendshoring and regionalization. These developments lead to a high-risk perception – 29% of respondents rank business interruption as a top peril, placing it at #3, although it drops a position year-on-year.

Unsurprisingly, political risks and violence climbs two places to #7, its highest-ever ranking. The closely linked risk of changes in legislation and regulation – which includes trade tariffs – ranks #4 globally, unchanged year-on-year but with an increase in respondents, driven by concerns over growing protectionism. In fact, global supply chain paralysis due to a geopolitical conflict ranks as the most plausible "black swan" scenario likely to materialize in the next five years, according to 51% of the respondents.

The full report is available at: 2026 Allianz Risk Barometer.

With inflation, a higher cost of living, strained budgets, and job market instability, today's consumers are more price-sensitive than ever. One of the key drivers of rising costs in consumer service industries, such as insurance, is abuse of the legal system. 

When the judicial process is unfairly manipulated to generate profit through excessive or unwarranted litigation, the financial fallout extends well beyond the courtroom. With insurance companies forced to spend more defending against these claims, losses are driven up and, therefore, premiums. The result is an inequitable system that places consumers, businesses and the broader U.S. economy at the forefront of absorbing these escalating costs, often out of the public eye but deeply felt in monthly insurance costs.

What is Legal System Abuse and Third-Party Funding?

Legal system abuse is the misuse of laws and judicial processes to increase litigation for profit, often through harassment, control or financial exhaustion and for purposes other than justice. When left unregulated, these practices can drive up costs for consumer services, such as insurance. Defending against these claims increases insurer losses, which in turn drives up premiums for policyholders.

Legal system abuse can include several facets, such as third-party litigation funding, fraudulent claims, exaggerated damages, and deliberately enacted practices to increase costs, delay settlements, and inflate verdicts. Third-party litigation funding is the broad term for providing money to a party to pursue a potential or filed lawsuit in return for a portion of the damages or settlement awarded by the court. Compensation of this nature is often provided by hedge funds, investment firms, individual investors or foreign entities. Since the third-party funders back most, if not all, of the legal expenses associated with litigation, plaintiffs have minimal risk in bringing their claims to court, whether merited or not. According to its 2024 Litigation Finance Report, litigation finance firm Westfleet Advisors disclosed $16 billion in third-party litigation funding (TPLF) assets under management.

This power dynamic puts pressure on insurance companies that are defending against these claims to settle them out of court to avoid trial expenses and drawn-out legal proceedings, even if the claims lack legitimacy.

How Does Legal System Abuse Affect Consumer Finances?

In November 2025, the Independent Insurance Agents & Brokers of America (the Big "I") released a national survey of consumers ages 25 and older who have home, auto or business insurance. It found that 64% of respondents were concerned that excessive lawsuits increase their insurance premiums. According to the survey, 81% believe that the legal system is used in ways that unfairly drive up insurance costs. In addition, eight in 10 (80%) also felt that their premiums would increase due to excessive lawsuits, even if they had never filed a claim themselves.

Consumers' concerns are well-founded. According to A Consumer Guide: How Legal System Abuse Impacts You, released in June 2025 by Triple-I and Munich Reinsurance America (Munich Re US), legal system abuse has resulted in $6,664 in added annual costs for the average American family of four. The excessive litigation prompted by the abuse has cost the U.S. economy 4.8 million jobs and $160 billion in annual tort costs for small businesses.

How Does Legal System Abuse Directly Affect Insurance Losses and Premiums?

Recent findings suggest that legal system abuse drives costs to rise above typical economic inflation rates, ultimately leading to higher prices for average insured Americans. In October 2025, the Insurance Information Institute (Triple-I) and Casualty Actuary Society (CAS) released a report showing that legal system abuse and related litigation trends drove liability insurance losses up by more than $230 billion over the last decade, a figure well over what can be attributed to economic inflation alone.

According to the report's key findings, legal system abuse, inflation-increased losses and defense and cost containment (DCC) for personal auto liability insurance increased by $91.6 billion to $102.3 billion. Commercial auto liability rose by $52 billion to $70.8 billion, property liability by $4.6 billion to $4.8 billion and other liability insurance by $83.4 billion to $103.3 billion (totaling $281.2 billion).

A separate Triple-I report published in September 2025 found that, from 2014 to 2023, an increase in motor vehicle tort lawsuits resulted in $42.8 billion in losses for insurance companies. Data from these findings underscore how litigation dynamics directly influence insurance losses. While the datasets vary, Triple-I's motor vehicle report supports the claim that roughly one-third of "increasing inflation" in auto vehicle insurance losses is due to legal system abuse. These actions, in tandem with mounting legal system pressures, are directly contributing to rising insurance premiums, meaning families are paying more for their coverage at the end of the day, regardless of whether they have filed a claim.

What Can Be Done to Fix an Unfair System?

According to the 2025 report from Big I, 88% of respondents stated that reducing unnecessary lawsuits is important for controlling insurance costs. In addition, 84% reported that they would support reforms if certain legal practices were making their insurance more expensive.

To do this, federal and state reform of tort law and third-party litigation funding must be implemented. But bringing about change to the judicial system is not an easy feat and requires the collaboration of many stakeholders working in tandem toward reform. When asked in the Big "I" survey, a majority (55%) of consumers agreed that the state and federal government should spearhead efforts to address the issue. Many respondents also pointed to insurance companies (34%) and courts (33%) as playing important roles in bringing about reform.

Legal system abuse is a growing financial concern that affects everyday consumers, businesses and the broader insurance industry. Without the necessary intervention to reform tort law, provide oversight of third-party financing, and raise public awareness of these practices, insurance costs will continue to rise, outpacing inflation and putting even more financial strain on American families. Recent data has revealed that consumers are signaling that they want a more just, accountable system. To do this, government officials at the state and federal levels will need to work hand-in-hand with insurance companies and the court systems to bring about change. By continuing to spread information about these practices and to advocate for balanced reform, we can help ease the financial burden on today's policyholders and bring greater transparency and justice to this system.

For decades, insurers have carried the unwelcome reputation of being technological laggards in the financial services sector.

Burdened by enormous, complex legacy systems that are costly to update, insurers have found it challenging to implement new technologies. Add that to cultures and regulations prioritizing strong, stable balance sheets over risky technology punts and insurers have often fallen behind adjacent industries in adopting innovation - or so the story goes.

One of the main reasons legacy systems have hamstrung insurers is that they were one of the first industries to adopt mainframe computers, directly leading to an unfortunate era of monolithic systems. More recently, the industry has been working to change this perception. In a period of rapid transformation, insurers have been quick to adopt new technologies like cloud computing and continue to drive cloud adoption across the value chain.

This reputation of being slow to evolve is especially unfair when applied to claims functions. Over the past two decades, claims operations have undergone significant evolution, shifting away from in-house mainframes, burdensome human-led decision making and inconsistent data capture toward more streamlined, digital-first processes. Insurers are much more data-led today, much more focused on how analytics and AI can improve the efficiency and cost of their operation, while, at the same time actually improving the customer experience.

Six key areas where claims functions can find value from AI:

1. Fraud detection

AI can analyze patterns and anomalies in data to identify potential fraud faster and more accurately than traditional methods. This not only reduces financial losses but also improves the overall integrity of the claims process. For example, AI can analyze structured and unstructured data to uncover hidden fraud patterns and complex relationships within social networks, substantially increasing fraud detection rates.

AI solutions can enhance fraud detection by using machine learning models that continually learn and adapt to emerging fraud patterns. It's important to note that we are not talking about one fraud model here. A network of fraud models working together makes the biggest difference. Insurers have significantly improved fraud detection by implementing a series of fraud scoring models, including supervised models and unsupervised neural networks, within their end-to-end insurance analytics and pricing platforms.

2. Claims triage and allocation

AI optimizes triage and allocation, ensuring each claim is handled by the most appropriate team or individual. This improves operational efficiency and ensures that complex claims receive the attention they need, while simpler claims are processed quickly. AI-driven triage can prioritize claims based on severity, complexity and potential cost, leading to better resource management and faster resolution times. This not only speeds up the claims process but also enhances the accuracy of assessments and decisions.

This is particularly important in markets like the United States, where legal representation in casualty lines is rising, driving up claim costs and closure timelines. Identifying claims that are likely to be represented or litigated early in the process helps triage the claim to the right team, enabling proactive settlement strategies and controlling costs.

3. Predictive analytics

AI-powered predictive analytics can forecast the likelihood of various outcomes, such as the probability of a claim being fraudulent or the potential cost of a claim. This allows insurers to make more informed decisions and allocate resources more effectively. Predictive models can assess the ultimate case value at the First Notification of Loss (FNOL) stage, enabling proactive claims handling and reducing indemnity costs.

Insurers often underestimate the value of powerful predictive claim models. This isn't just about using better prediction in claims, for example, where we have helped insurers route claims that are likely to jump to specialist teams using unstructured data. Insurers need to be thinking more holistically and using the insights their predictive claims models offer in reserving and pricing processes.

4. Customer experience

Enhance the customer experience by providing faster and more accurate service. AI-powered chatbots and virtual assistants can handle routine inquiries and guide customers through the claims process, while AI-driven decision support tools help claims handlers provide timely and accurate responses.

The result? Higher satisfaction and loyalty, as evidenced by NPS scores, and stronger retention.

5. Cost optimization

Automation of routine tasks and improved claims processing accuracy reduce operational costs, freeing resources for high-value activities. For instance, AI can optimize repairer selection based on core claim metrics, leading to cost-effective repairs and shorter settlement times.

AI-driven automation can handle repetitive tasks such as data retrieval, input processing and quality checks, resulting in significant time savings and improved accuracy.

In casualty lines, an integrated generative AI tool can help summarize legal correspondence, medical reports and investigation updates and recommend negotiation strategies, saving claims handlers significant time.

6. Real-time decision support

Real-time AI engines provide claims handlers with instant insights for faster decisions. This includes assessing the likelihood of fraud, determining the best course of action for a claim, and identifying opportunities for cost savings.

These engines can integrate multiple models to provide comprehensive insights and support across the claims process. This real-time capability ensures that claims are processed efficiently and accurately, reducing delays and improving overall performance.

Insurers see enormous benefit by deploying scoring models that can feature in their claims systems, so claims handlers have greater real-time insight when making decisions.

What's next for AI and claims innovation?

The integration of AI into the claims process offers insurers numerous benefits. By leveraging AI, insurers can transform their operations, delivering better outcomes for both their business and their customers. But insurers that are tempted to lag might well be targeted by fraudsters or see their customers turn away due to relatively slow processing and decisioning.

The reality is that the scope and breadth of claims AI applications are vast, and part of the secret to successful AI deployment is to identify and prioritize effort.

One of the continuing and increasing challenges in clinical and cost modeling is translating scientific advances into real-world practice at scale. Years can pass before new evidence meaningfully influences care delivery, benefit design or financial planning that affects insurance premiums. Closing this gap between what is known and what is applied has proven difficult across the healthcare ecosystem.

This is largely the result of medical knowledge that is not inherently computable, which limits precision, transparency and scalability across the healthcare ecosystem. Making medical evidence usable in real-world insurance coverage decision-making requires computational approaches that bridge medical science, clinical practice and economics.

With medical knowledge becoming computational, a new class of solutions is emerging – one that connects the science of medicine with the economics of delivering care and managing risk. This approach structures evidence-based clinical knowledge in a form that can be reasoned over transparently, helping organizations compress the knowledge-to-practice cycle and make more informed decisions under uncertain conditions.

At its core, this methodology supports better risk stratification and management by grounding prediction in clinical understanding. Rather than relying solely on historical usage patterns, organizations can now evaluate patient journeys, assess plausible future trajectories and reason about clinical and financial risk with greater clarity.

Aligning Clinical and Financial Perspectives

Most healthcare in the United States is employer-driven and sits at the intersection of clinical insight, economics and access. Yet these components often remain siloed. Clinical information, claims data and financial models are rarely aligned in a way that supports coherent and holistic risk management.

Risk-bearing organizations routinely navigate clinical and financial decisions that are not intrinsically connected. In the absence of alignment between these perspectives, early risk identification and confident action are challenging.

Introducing a computational layer that connects medical evidence with real-world data helps bridge this divide. Clinical guidelines, care pathways and research are translated into explainable models of clinical logic. When an individual's health history is evaluated against this foundation, organizations gain a more complete and interpretable view of risk.

Instead of a standalone risk score, this approach offers a transparent, evidence-grounded view of risk that informs pricing, underwriting, budgeting, care management and more.

Explainability as a Requirement

Explainability also plays a central role in whether AI can be trusted in healthcare risk management. Decision makers must be able to see how a conclusion was reached, how evidence was connected and why certain outcomes are considered plausible.

When models reflect real clinical reasoning and make that reasoning transparent, they become usable across teams. Actuaries, care managers and leadership can operate from a shared understanding rather than interpreting disconnected outputs.

Research increasingly highlights the importance of interpretable models that align with clinical practice. Predictions that cannot be examined or explained offer limited value in environments where financial and human outcomes are closely intertwined.

A More Precise View of the Future

One of the key advantages of clinical modeling is its focus on individual trajectories rather than broad population categories. A diagnosis alone does not indicate whether a condition is stable or worsening. A procedure does not explain whether it reflects appropriate care or avoidable deterioration. Individuals with similar claims histories may face very different futures.

When these distinctions are made visible to all, organizations can act earlier and with greater confidence. This enables targeted intervention, education or more effective planning, driven by understanding and contemplation rather than hindsight.

This clarity helps align clinical and financial teams. Clinical experts understand how health evolves; financial teams understand how cost behaves. When both are connected through a shared, evidence-based model, organizations can make more confident decisions around pricing, benefit design and care management investment. This shared foundation reduces friction between teams by grounding discussions in the same clinical and economic context.

Moving Forward Responsibly

As AI adoption accelerates in healthcare, responsible use remains essential. Models must address bias, protect privacy and preserve meaningful human oversight. Clinical modeling does not replace professional judgment – it augments it by providing a clearer, evidence-grounded view of uncertainty and risk.

When prediction is grounded in clinical understanding, risk becomes more visible and more manageable. Organizations can see not only what may happen, but why, enabling more responsible action.

By transforming medical evidence into computational knowledge and applying AI to that foundation, this approach enables more transparent, aligned and effective risk management – benefiting patients, employers, insurers and the broader healthcare ecosystem.