ITIC Global 2026 | Technology and governance
Technology is driving change in the insurance sector. David Phipson, Russell Corbould-Warren, and Delphine Icart discussed with Marc Banting, COO, ITIC, how the developments in AI and other advanced technology are affecting clients and customers and how good governance is needed throughout the introduction of these technologies
The ITIJ team are reporting from ITIC Global in Istanbul this week (1–5 November 2026), sharing the discussions that take place at the conference. Read all reports.
David Phipson, Managing Director and Head of AI, Lodestone Solutions
Phipson explained how artificial intelligence (AI) is used across the entire travel claims journey; describing how it is used in claim lodgement, in claim assessment, and in client communications. As AI is a powerful tool, Phipson added that there have to be strict levels of governance and controls. He pointed out three key areas that AI has to pass through in order to be used in during the claims journey: technical accuracy tests, human tests and reviews, and having a human in the process loop: “Each step of the process we’ve had to test and validate using analytics.”
The results of AI on productivity were significant over one year. Phipson stated that the time saved per claim was 75% less, and that it was able to replace 50% of the claims workforce. Phipson said: “AI is very good at summarising documents, reading all the documents, and identifying where that human needs to focus, and pointing them to problematic areas in the claim. Then they are much better able to formulate and come to a decision-making decision on the claim and move on.”
Each step of the process we’ve had to test and validate using analytics
By using AI in insurance claims, Phipson said that there were lessons that had been learned, which broke down into when it was best to use AI, best to use code, and best to use humans.
Phipson concluded by discussing where AI would be in the future, pointing to a continued rollout in the next half a decade, and the efficiency that can be made in claims. He also mentioned that it would be involved in emergency assistance, and that, more broadly speaking, there would be more strategic thinking and reactive ability, taking over more critical actions, following the developments seen in the increasing use and ability of self-driving cars.
“You need to get on the AI wave, get on the surfboard, ride the wave, otherwise you’ll get dumped. Overall it’s a positive message. There is a lot of friction and AI can ameliorate that process. And we should take advantage of it,” concluded Phipson.
In response to a question from the audience whether AI could recognise if there had been overcharging or overtreatment in an insurance claim. Phipson said that AI was already doing this and gave an example where AI would identify immediately a discrepancy if a Canadian consultation was charged at $600 but should be $150.
Russell Corbould-Warren, Senior Vice President of Insurance, Cover Genius
Corbould-Warren said that only 30% of shoppers had bought protection at an e-commerce checkout, but also that “69% of those same shoppers value protection at checkout”, highlighting the disparity as a real demand for this service, equating to an opportunity for insurers.
Corbould-Warren explained that the online checkout had become the new shopfront, where industries such as travel, health, mobility, retail, and fintech all can offer protection. He stated that the global embedded insurance market was worth US$188.5 billion in 2026 with an estimated growth to $1.24 trillion by 2033.
The global embedded insurance market was worth US$188.5 billion in 2026 with an estimated growth to $1.24 trillion by 2033
The development of AI agents means that they are starting to shop for customers. Corbould-Warren said that agents could find offers and present them directly to the customer. He showed a simplistic example of an agentic conversation, where the consumer asked the AI agent to find then book a trip, add insurance, and then amend and add extra cover to that insurance.
He stressed that when AI suggests cover, the final decision still goes to the human customer to make. Corbould-Warren supplied a case study of his own experience when buying some shoes for his son, where this process occurred quickly and easily. He said that 78% of people were open to AI-assisted buying, but that 95% of travellers wanted to keep a final human decision on whether or not to approve the protection.
Corbould-Warren concluded that agentic embedded insurance is now available and “in the last year has really accelerated”.
When asked by moderator Marc Banting about who was responsible for the accuracy of the product presented by AI, Corbould-Warren said: “It’s our licensing responsibility, to make sure that the distribution channel we’re using, whether it’s a more traditional path or an agentic model, has enough information provided, that I’m comfortable as the licence holder, and that the customer understands.”
Delphine Icart, Chief Executive Officer, Global Health
Icart began by explaining what international private medical insurance (IPMI) is and who it is for, identifying that people who “live, work, study, or travel across borders” are in need of private healthcare in multiple countries. She said that IPMI products need to achieve four main requirements: portability, international servicing, global claims management, and global compliance. She said that because IPMI exists in an area where “healthcare insurance, technology, and regulation intersect”, good governance is important, and posited whether technology is helpful or a hindrance.
Icart pointed out that technology helps with providing insurance products when asking: “How do you make it easy for the clients while managing the complexity in the background?” She then said that using technology wasn’t a matter of just automating everything that can be automated, but that a deeper consideration should be made about which decisions and functions ought to be delegated. She added that although efficiency is important in health insurance, the value of context, fairness, compliance, and accountability is also hugely significant.
The effectiveness of governance is greater than making it compliant, and safe and secure, argued Icart. She said that it needed to focus on real outcomes, and that there were four questions that needed to be answered in order for digital insurance to be considered appropriate: is the outcome fair, can we explain the decision, who is accountable, and can we demonstrate control?
Governance of digital insurance should not be the final stage of the transformation process but it should instead be a consideration and part of the architecture from the beginning
Regarding going from innovation to trust, Icart said: “I really believe that we can get there.” Governance of digital insurance should not be the final stage of the transformation process, said Icart, but it should instead be a consideration and “part of the architecture from the beginning”. She said that there were five parts to this development from design, through approval, deployment, and monitoring, leading to intervention with corrections and learning. She added that clear guardrails to the implementation of the transformation allow for confidence in the process and the ability to experiment and scale the project responsibly.
A question from the audience related to distrust of giving AI medical information. Phipson said that it was “a change management exercise and trust will come over time”. Icart said that as an ecosystem, insurers were finding the balance of getting enough information in order to help provide healthcare quicker with guaranteed access.