The AI arms race and the Strategic, Financial, Operational and Regulatory Risks that it creates
Keynote presentation by Simon Torrance
Why the world needs new AI risk solutions — and how insurers can lead the way
The Real Risk Isn’t What You Think
When most people talk about AI risk, they think of cyberattacks, rogue algorithms, or governance failures. And yes, those are real threats.
But according to Simon Torrance, founder of AI-Risk, the biggest risk is much simpler — and far more urgent:
Being outcompeted by companies who use AI better than you.
For the insurance industry, that’s both a warning and a significant opportunity.
A Tornado Is Coming
AI moves fast. It’s unpredictable. It doesn’t behave like traditional software. It learns, adapts, and evolves without human instruction.
Simon compares it to a tornado: fast-moving, powerful, and incredibly hard to contain.
AI isn’t just a tool. It’s a new capability that can transform business models, industries, and societies. The problem is — very few are prepared for its impact.
AI Risk: The Four Categories
Through a think tank with Marsh and other major insurers, Simon and his team identified four core types of AI risk:
1. Strategic Risk
Falling behind competitors who implement AI more effectively.
2. Financial Risk
Wasting capital on failed or low-return AI investments.
3. Operational Risk
Exposure to new cyberattack vectors, talent shortages, and cascading infrastructure failures.
4. Regulatory Risk
Increased liability due to non-compliance with emerging AI laws like the EU AI Act.
These categories cover the full spectrum of what AI disruption looks like — and most insurers are only beginning to address them.
Don’t Repeat the Cyber Mistake
Cyber risk caught the industry off guard. The result?
A $3 trillion global protection gap — with insurers covering only a fraction of the actual loss exposure.
Simon warns we are heading toward a repeat scenario with AI. The world is racing to adopt it. The industry is slow to respond.
AI Is Also a Growth Market
This isn’t just about defence — it’s about innovation.
Insurers have a chance to lead by:
Offering AI-specific risk advisory and coverage to enterprise clients
Building AI governance tools and services
Developing prevention, prediction, and mitigation platforms
The enterprise and SME market is the best place to start. Adoption is high. The need for help is real.
Why AI Risk Is So Complex
Unlike traditional systems, AI is non-deterministic. It:
Adapts without instruction
Makes autonomous decisions
Learns and evolves over time
Scales across environments invisibly
These qualities break traditional insurance logic — especially around foreseeability and responsibility.
For underwriters and actuaries, AI raises uncomfortable questions: Who’s liable? Who controls the output? How do you quantify risk from autonomous agents?
A New Kind of Turing Test
Simon references Mustafa Suleyman, co-founder of DeepMind, who recently proposed a modern Turing Test:
Give an AI $100,000 and see if it can turn it into $1 million — autonomously.
The AI must identify a market gap, design a product, manufacture it, and sell it online — all without human control.
This may sound like science fiction. But leading AI experts believe it’s less than two years away.
Building the First AI Risk Taxonomy
Through the AI-Risk think tank, Simon’s team created the first taxonomy of AI-related risks — over 80 independent risk categories spanning technical, operational, political, and economic areas.
Key findings:
Over 79% qualify as “high-risk” under the EU’s AI definition
68% occur with high frequency
Most companies are not tracking these risks — let alone managing them
The most severe risks aren’t always obvious. The real threat comes from unintended outcomes: biased AI hiring decisions, model drift in healthcare, or widespread infrastructure failures from AI-driven automation.
So What Should the Insurance Industry Do?
Simon outlines a four-part roadmap for insurers:
1. Educate senior leaders
AI risk must be understood at the executive level — not buried in compliance or IT.
2. Build joined-up AI strategies
Many companies are running fragmented AI experiments without a strategic framework.
3. Avoid silent AI exposure
Just like cyber, AI-related claims will emerge under policies not designed for them.
4. Build more than coverage
Insurers should offer education, assessment, prevention, and mitigation solutions — not just indemnity.
Final Thoughts: A Call to Innovate
AI is coming. Fast. And it will change everything.
Simon ends with a quote from the UK’s AI Safety Summit:
“We agree on the urgency behind understanding the risks of AI — to ensure the long-term future of our children and grandchildren.”
The insurance industry has a once-in-a-generation opportunity to help the world adopt AI safely — and grow in the process.
The question is: who will lead, and who will get left behind?
Explore the full AI Risk taxonomy or request a board-level briefing:
Visit ai-risk.co or connect with Simon Torrance on LinkedIn
The Shift Has Already Started. The Winners Are Scaling Agentic Teams Now
The Agentic Enterprise Blueprint maps your transformation from pilot purgatory to exponential scale.
