In a recent interview with Reinsurance News, Andrew Johnston, author of the InsurTech Quarterly report series and Global Head of InsurTech at reinsurance broker Gallagher Re, discussed how investment trends, artificial intelligence, and emerging liability risks are reshaping the insurance and reinsurance market.
To start, Johnston highlighted a clear shift in funding patterns compared with recent years, suggesting a sustained increase in baseline investment levels, rather than volatility driven by a handful of large transactions.
“We’re now seeing the second consecutive quarter of funding above $1.6 billion, which is a dramatic uptick in what we saw for the prior two and a half years, where funding was incredibly consistently in and around the $1.1 billion range. And where it deviated from that $1.1 billion, it was because one or two individual, very large deals were being done,” said Johnston. “What we’re seeing is the beginning of a new consistent funding level that is $500 million more than what we’ve witnessed in the prior two years.”
When asked about where funding is flowing, Johnston was unequivocal about the dominance of artificial intelligence. He commented: “By a mile AI technology, about 95% of funding is going into technologies that identify as having AI tools.” He explained that much of this capital is being directed towards tools built around machine learning and large language models, particularly those capable of analysing large volumes of data and returning human-like outputs.
He described these capabilities in detail: “It’s things that could process data very quickly or it’s things that can run a model of something that’s very predictable. When we’re talking about AI technologies, we’re talking about tools that are mimicking human cognition, and so it’s generally tools that can consume a lot of data and then give you a human like analysis.”
He explained that these systems are becoming more integrated into everyday work processes: “For instance, with some AI tools, you might upload an 800-page claims file, provide the individual’s name, reference number and details, and it will return the relevant information, almost like speaking with someone.”
Fraud detection is one of the most prominent use cases attracting investment. Johnston said: “We’re seeing a lot of tools that can verify image authenticity. There are instances where people are faking car crash images or hail damage images.” He further explained how models identify suspicious patterns: “They’ve been trained to spot red herrings, historically, we know that if these data points all happened together, it’s 70% more likely to be a fraudulent claim. Those claims then get highlighted as suspicious and then reviewed by a human being.
“So, any technology in and around that space is collecting a lot of attention, and therefore cash dollars in form of investment.”
Turning to the development of AI liability insurance, Johnston described a market that is quickly taking shape across the value chain. “It’s certainly developing. Now we have a burgeoning suite of market players, from the consultants, the analysts, to the MGAs that are creating products for the AI liability space, that themselves come with a lot of expertise, all the way through to reinsurance markets who are prepared to support these MGAs and insurance companies that are offering these products.”
He drew a direct comparison with the early evolution of cyber insurance. “In terms of its trajectory as a theme, it’s very similar to what we saw in cyber about 10 years ago. This emerging risk class that was noticing losses that were coming from the digital space, that you could argue, sat in a traditional loss pillar, and yet clearly wasn’t.” He added: “I think we’re going through exactly the same thing with AI liability.”
As AI adoption accelerates, Johnston emphasised that new forms of risk are emerging, particularly around accountability. “Transparency in the value chain is key. If I’m buying a service from you and that service goes wrong, but you delegated what you did to an AI functionality, do you know where the issue began? Where should the transparency be?”
He suggested regulatory expectations could evolve, explaining that there may come a day where regulators force companies to share training manuals and modules with the general public, so people know exactly what sentiment analysis and what bias has been done.
He also stressed that the market is still at an early stage in addressing these issues. “We are at that kind of embryonic phase. I think the best thing we can do is just communicate and be explicit.”
On how insurers and reinsurers should respond, Johnston encouraged active engagement with the space. “Be open to the space. Be open to being educated, and open to speaking to companies that are being very proactive.” He added that firms should rely on their existing expertise: “My advice to reinsurers when there’s a new class of business, is always to be brave and trust their underwriting instincts, their portfolio management instincts.”
He concluded that the opportunity is likely to expand significantly. “I don’t really see how they can miss out on the opportunity, because I think the opportunity is going to continue to grow, and there’s going to be enough for everybody to see first-hand the opportunities as they come through.
“We’re seeing a lot of reinsurers be very proactive and have dedicated members of staff that are talking to their own underwriters about how this could impact their own businesses.”
So, overall, the sector is entering a phase characterised by stronger and more consistent investment, rapid AI adoption, and the formation of new liability frameworks that echo the early development of cyber insurance.
