Europe’s AI Reckoning Is an Execution Problem

By Adrian Mincher, Director at Earnix

Europe has spent the past year debating AI regulation as if the defining challenge is how to slow AI down safely. In reality, most insurers face a different problem: they cannot operationalise AI fast enough across pricing, underwriting, claims and customer engagement to keep pace with risk, customer expectations, or market volatility.

That tension is becoming more visible across the UK and Europe as regulators accelerate efforts around governance, transparency and accountability. Delays and revisions to the EU AI Act may dominate headlines, but the real story inside insurance boardrooms is more pragmatic. They are asking why so much investment still struggles to translate into enterprise-scale results.

Earnix’s 2026 Trends research, based on insights from more than 400 insurance professionals globally, including UK and European carriers, reveals an industry actively embedding AI into workflows while confronting with the realities of scaling it responsibly.

AI adoption is accelerating across the market, yet execution remains uneven.

In the UK, 55% of insurers say AI is already integrated into some functions, compared with just 38% claiming it is fully integrated across most functions. That distinction matters. “The industry has moved beyond experimentation, but many insurers are still stuck between adoption and execution: AI exists in pockets, but not yet at the scale, connectivity or velocity needed to transform the business.

Generative AI adoption tells a similar story. UK insurers are moving aggressively into workflow-adjacent use cases such as unstructured data processing, quote generation and customer profiling. These are sensible starting points because they deliver fast time-to-value without forcing wholesale replacement of legacy systems. That distinction is becoming central to how insurers evaluate the next generation of AI infrastructure.

The execution gap is where many AI strategies begin to fracture. The problem is rarely ambition. It is operationalisation. Insurers are discovering that scaling AI requires far more than deploying models. It demands data interoperability, governance frameworks, workflow redesign, explainability, auditability and organisational alignment. That level of infrastructure is difficult to build inside businesses still operating across fragmented technology estates and siloed decision-making structures.

The survey findings make this clear. Data quality concerns remain exceptionally high, with 81% of UK insurers worried about the integrity of AI inputs. Meanwhile, 91% plan to increase investment in third-party data to strengthen decisioning quality.

AI performance now depends on the decisioning architecture around it. Better models alone will not solve execution latency.

This is particularly important in Europe, where AI regulation is converging with broader expectations around Consumer Duty, transparency and accountable automation. UK insurers already frame AI ethics primarily through the lens of regulatory and legal exposure rather than abstract ethical debate. That creates a fundamentally different operating environment from Silicon Valley’s ‘move fast and break things’ mentality.

Insurance cannot afford improvised automation. The winners will be insurers that connect governance, pricing, underwriting, claims, customer engagement and AI decisioning into one operating model.

That is why the conversation is shifting from AI as a tool to AI as an operating system for insurance decisioning.The next phase of insurance AI will be defined by platforms that turn insight into governed, explainable action at enterprise scale. This is the insurance innovation crisis: not a shortage of ideas, but a widening gap between ambition and execution.

Europe may be moving more cautiously than some markets. But insurance has always been a business where trust matters more than hype. And in insurance, the real AI advantage will belong to those that can execute with speed, governance and trust.