INTERVIEW: Can Insurance Become the Trust Engine for AI and Digital Assets?

In this interview with Impact Newswire, Christian Capes-Davies, Global Head of Distribution and Innovation at Relm Insurance, explains how the rapid convergence of artificial intelligence, blockchain and digital assets is reshaping financial risk and forcing insurers to rethink decades-old underwriting models. As AI agents begin executing payments, approving financial transactions and making autonomous decisions with limited human oversight, questions that once belonged to technology developers are increasingly becoming insurance questions: Who is liable when an AI makes a costly mistake? How should insurers assess algorithms that continuously learn and evolve? And can insurance become a trust signal for institutions adopting AI-powered financial products?

Artificial intelligence is rapidly moving beyond chatbots and recommendation engines into systems that can execute financial transactions, approve loans, detect fraud and manage investment portfolios with minimal human intervention. Technology companies, banks and payment firms are increasingly developing AI agents capable of making autonomous financial decisions, raising new questions about liability, operational risk and insurance. If an AI agent authorizes a fraudulent payment, misprices an asset or makes an erroneous investment decision, legal responsibility may no longer rest with a single party but could be shared among software developers, financial institutions, customers and insurers.

The shift comes as investment in AI continues to accelerate. Global private AI investment reached $252.3 billion in 2024, while generative AI alone attracted $33.9 billion, according to the latest AI Index Report by Stanford University’s Institute for Human-Centered Artificial Intelligence. Financial services has emerged as one of the fastest-growing sectors for enterprise AI adoption, with banks increasingly deploying AI for fraud detection, credit underwriting, compliance monitoring, customer service and algorithmic trading. Industry analysts estimate that AI could contribute hundreds of billions of dollars annually in productivity gains across banking and capital markets over the next decade.

At the same time, digital assets are becoming increasingly integrated into mainstream finance. The global stablecoin market has grown to more than $250 billion in circulation in 2025, while banks, asset managers and payment companies are accelerating efforts to tokenize deposits, bonds, money market funds and other real-world assets. Consulting firms project that tokenized assets could reach several trillions of dollars by the end of the decade, fundamentally changing how securities are issued, traded and settled. These developments are creating entirely new categories of financial, operational and cyber risk that traditional insurance products were never designed to cover.

Insurers are also confronting a rapidly evolving threat landscape. Cybercriminals are using generative AI to produce more sophisticated phishing campaigns, malware and identity fraud, while organizations increasingly worry about AI systems making incorrect autonomous decisions, producing biased outcomes or suffering model failures that disrupt business operations. At the same time, financial institutions face growing risks from smart contract vulnerabilities, blockchain infrastructure failures and interconnected AI systems capable of executing transactions at machine speed, potentially amplifying losses before humans can intervene.

Against this backdrop, insurers are expanding beyond conventional cyber coverage to evaluate risks associated with autonomous AI systems, digital assets and decentralized financial infrastructure. As underwriting increasingly depends on assessing algorithms alongside human behavior, the insurance industry is expected to play a growing role in determining which AI-powered financial technologies earn institutional trust and achieve widespread adoption.

Christian Capes-Davies, Global Head of Distribution and Innovation at Bermuda-based Relm Insurance spoke with editor Faustine Ngila on how insurers are adapting to autonomous finance, the lessons learned from underwriting the crypto industry, and why insurance may become a critical enabler of the next generation of AI-driven financial services.

1. Stablecoins and tokenized assets are moving into mainstream finance. How are these technologies changing the types of risks insurers are being asked to underwrite?

The big thing with stablecoins is that it depends what type of stablecoin you are talking about. We have already seen one of the biggest crypto failures in history come from a stablecoin that could not maintain its peg. Luna / Terra is the obvious example. That was an algorithmic stablecoin, and when confidence broke, it created something that looked very similar to a run on a currency. You can almost think of it like the George Soros attack on the British pound: once the market no longer believed the peg could hold, the structure came under enormous pressure.That is very different from a fully backed stablecoin like USDT or USDC, where the

reserves are intended to sit one-to-one, or in some cases over-collateralized, against recognized financial instruments. So from an insurance perspective, you have to understand what is actually behind the stablecoin. Who is issuing it? What assets are backing it? Is it cash, Treasuries, or something else? What are the liquidity provisions? Can it be redeemed under stress? You also have to look at third-party dependency. A company may launch a white-labeled stablecoin, but the underlying technology, custody, or reserve infrastructure may sit with someone else. So you are no longer underwriting only the token itself. You are underwriting the issuer, the technology provider, the collateral, the liquidity model, and the stability of the whole structure.

2. As AI agents begin executing payments and making financial decisions autonomously, who ultimately bears liability when something goes wrong: the developer, the financial institution, the customer, or the insurer?

It is a huge topic, and the honest answer is that it comes down to contract and control. An AI agent does what it is instructed to do, but it can also take steps within its guidelines that were not necessarily anticipated by the person deploying it. If the fences around that agent are very tight, it can be incredibly powerful. If they are not, that creates a very different risk profile. So the liability question is: who built it, who deployed it, who set the parameters, and who was responsible for checking what it was doing? It might sit with the technology provider. It might sit with the bank or financial institution using it. It might sit with the company that deployed it into the customer journey. It is definitely not that the insurer becomes liable for the agent’s decision. The insurer may have to pay a covered claim depending on the policy and the facts, but the legal liability sits with the party responsible for the error, the deployment, or the failure of oversight.That is why contract language, auditability, and governance become so important when AI agents start acting inside financial systems.

3. There has been a lot of media focus on AI-driven cyberattacks. Where do you think the insurance conversation around AI needs to move beyond cyber?

AI-enhanced cyberattacks are absolutely a real concern. AI can make a normal person more capable, and it can make a sophisticated attacker much faster. You could have attacks that used to take months being planned or executed in days, but I don’t think the insurance market should only be looking at the cyber angle.

The bigger issue is that AI is starting to touch ordinary business processes everywhere. Manufacturing, healthcare, legal services, engineering, finance, architecture. These are not always the flashy examples, but they may be where the real risk sits. Think about a lawyer using AI and relying on false cases. Is that a professional negligence issue, or is it an AI liability issue? Think about an engineer or manufacturer using AI to make a design decision. If that design fails, is that product liability, general liability, technology E&O, or something else?That is where the conversation needs to move. It is not only about AI being used to attack systems. It is about AI being used inside normal professional and operational decision-making, and how insurance policies respond when that judgment is wrong.

4. Traditional insurance models were designed around human decision-making. What changes are needed to insure financial systems where AI agents increasingly act independently?

The focus has to move from the decision itself to how the AI was told to make that decision. With a person, you can look at training, experience, professional standards, and judgment. With AI, you need to look at the framework around the system. How was it built? How was it tested? Who is checking the outputs? How often are they checking them? What happens when the AI makes a decision that was not expected? That is really where underwriting has to go deeper. It is not enough to know that a company is ‘using AI.’ You need to understand where it sits in the business, what decisions it is influencing, and what audit function exists around it. For a financial services company, that might mean AI touching payments, onboarding, fraud detection, compliance, credit decisions, or customer communication. Each of those creates a different risk. So the question becomes: does the company actually understand how the AI is operating, or has it just deployed it and hoped the outputs are right?

5. As banks and payment companies integrate blockchain, tokenization, and AI, which emerging risks concern insurers the most over the next five years?

The biggest concern is the combination of these technologies. Blockchain, tokenization, and AI are all complex on their own. When you start combining them inside payment companies, banks, or financial platforms, you create risk that is harder to isolate. AI is probably the biggest one because it enhances capability. It can help companies automate, scale, and make faster decisions. But it can also help bad actors move faster, find vulnerabilities, and create more sophisticated attacks.Then you have the more basic but very important questions. What is the technology stack? Who are the third-party providers? Where does the liability sit? What happens if the output is wrong? What happens if a company deploys something without fully understanding the consequences? That is the bit that should concern insurers. Companies are moving quickly, but speed does not remove liability. If anything, it makes understanding the decision process behind the technology even more important.

6. Can insurance become a competitive advantage for companies building AI- powered financial products, particularly when it comes to attracting institutional clients and earning consumer trust?

Yes, particularly with institutional clients.Consumers are often willing to use new technology without thinking too deeply about the risk. Institutions are different. They don’t want to inherit liability that they do not understand. For companies building AI-powered financial products, good insurance can be a real advantage because it shows that someone has looked under the hood. It shows that the company has gone through underwriting, that the controls have been tested, and that there is a mechanism to respond if something goes wrong. It is not a substitute for governance, controls, or good technology but it can sit alongside those things and give institutional clients more confidence.That is how insurance can help companies scale. It gives partners, investors, and customers more comfort that the risk is understood and that the company is serious enough to be underwritten.

7. Drawing on Relm’s experience underwriting digital assets and other frontier technologies, what lessons from the crypto industry’s evolution are most applicable to today’s rapid adoption of AI in finance?

The main lesson from crypto is that insurance is an enabler of business. In crypto, we saw a huge amount of capital, innovation, and ambition, but also major failures, losses, lawsuits, and regulatory change. AI is similar in the sense that there is a lot of money being deployed, the technology is moving quickly, and we do not yet know where all the risks will emerge.The lesson is that insurers cannot rely only on historical data, because the historical data often does not exist. You have to be close to the companies building the technology. You have to understand their business models, how they are regulated, how they are lobbying, where litigation might emerge, and how the risk is changing. That is the way Relm has approached digital assets, and it’s a similar principle for AI. You try to stay ahead of the curve by staying close to the clients and the industries themselves.

8. Beyond regulation, what role do you see insurers playing in determining which AI-driven financial technologies reach mainstream adoption and which struggle to gain market trust?

The truth is that traditional insurance can be a bit of a gatekeeper for emerging industries, whether the market likes that role or not. If a company cannot get insurance, it can be harder for institutions, commercial partners, customers, and investors to adopt what they are building. That’s one of the reasons Relm exists. Traditional insurance markets often struggle to understand frontier technologies, and that can slow adoption even when the underlying company is building something credible. But that does not mean every company should be insured. There are some businesses

where the model does not make sense, the controls are not strong enough, or the risk is too far outside appetite. In those cases, the inability to get insurance can tell the market something. Where the fundamentals are strong, though, insurance can help validate the business. It can show that the risk has been understood, that the company has been through a proper underwriting process, and that there is confidence behind the deployment. So insurers are not regulators, but they do influence trust. They can help responsible companies reach the mainstream, and they can also highlight where a risk is not yet ready for broad adoption.

Stay ahead of the stories shaping our world. Subscribe to Impact Newswire for timely, curated insights on global tech, business, and innovation all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Pitch it to us at info@impactnews-wire.com and reach the right audience worldwide