2026 - a gaze into the crystal ball of D&O

As 2025 draws to a close, it's time to not only reflect on the year just passed, but look ahead to the future. Here's three key predicitons for 2026.

1. Soft market cycle turns

The soft market won’t come to an end in 2026, but we are likely approaching the bottom of the cycle, and it’s time to start thinking about the climb back up. Rate adequacy has deteriorated rapidly since 2022, and now it’s time to pay the piper. 

Major names are reporting an uptick in claims, driven predominantly by rising EPL matters. Consecutive government budgets have placed continued pressure on SMEs, squeezing almost every sector and prompting warnings around the ability to hire and retain workers. With the Employment Rights Bill looming, insurers have few reasons to be optimistic about claims expectations. 

This is before even considering the continually near or record insolvency rates, the continued increase in AI-related and broader cyber exposures (significant enough even that some US carriers have already sought approval to apply AI exclusions for certain policies). Add in those other buzzwords: Geopolitics, supply-chain disruption, and emerging exposures such as the Economic Crime and Corporate Transparency Act, and it becomes clear why downward pressure on rate is increasingly unsustainable.

Lloyd’s has issued warnings about rapid softening across multiple lines, particularly Cyber and Casualty. D&O likely sits in the same bucket. And let’s not forget we are scarcely two years from the 2023 remarks by Lloyds calling the D&O market “moronic,” and the fact is that conditions have arguably worsened since then.

All in all, something has to give. Then again, I said most of this last year as well…

2. Cyber and D&O will appear increasingly blended in more mainstream policies

   
   

Most SME cyber policies now include some form of “management liability” extension. Usually sub-limited and certainly not as broad as a standalone D&O policy, these extensions still demonstrate the clear and growing exposure that cyber incidents create for directors and officers.

Beazley, one of the dominant players in cyber insurance, has already signalled an intent in the US to pull back from what it considers a rate-inadequate market. Cyber claims are not slowing however, and if that market hardens, pressure on D&O policies to respond to cyber-related matters will only intensify (particularly if D&O rates remain inadequate too).

There also remains a notable lack of awareness about how intertwined these products already are… and how much more integrated they will likely become. Brokers who treat Cyber and D&O as siloed placements, with limited cross-diligence, will increasingly fall behind those adopting a more holistic placement strategy. The same applies to insurers: maintaining two distinct product lines that respond to overlapping exposures is becoming inefficient and, ultimately, untenable. Insurers who can bundle Cyber and D&O seamlessly will win out in unfavourable trading conditions against those who can’t. 

3. AI will continue to disrupt

   
   

 AI is good (there, I said it). It’s great, even… when used in the right context. As a tool to support and inform, it is enormously powerful. But we should not leave it in charge of anything. Yet people are, and from an insurance perspective this is a disaster waiting to happen. When unmanaged AI makes a mistake, who is accountable? Ultimately, the buck stops with senior leadership, and guess who pays for that (yes, it’s the D&O carrier… probably).

As AI adoption accelerates, underwriting focus will have to shift. Insurers will have to scrutinise policyholders on AI. What models are being used? How are they governed? How do companies validate outputs? What tasks are automated — and is AI replacing people or empowering them? Is it making decisions or informing decision-makers? If you asked most insurers how many risks on their books rely on AI in core processes, very few could answer with any real accuracy.

Imagine a fundamental flaw in the reasoning for the next version of ChatGPT or Copilot.Now imagine 20% of your insured’s deploy a decision making tool on these models.. That’s an aggregation risk hardly any D&O insurer is considering. 

We will also start to see which “insurtechs” can survive mounting claims pressure. This is the moment where the wheat separates from the chaff. The abundance of capacity (and private equity) poured into “AI- and data-driven” MGAs and brokers over the past half decade will finally be tested by the fundamental question: was any of it genuinely better than what came before? Some will be. The true innovators, the ones who have done things differently, they’ll survive; but some won’t. Smoke and mirrors will be exposed. The difference between genuine insight and a dressed-up Excel pivot table will become painfully clear. 

AI isn’t going anywhere; the challenge is learning to harness it properly.