AI adoption in the insurance industry is growing at a pace that few could have predicted.
And along with this rapid advancement comes a concern that simply cannot be ignored: governance frameworks just aren’t keeping up with that speed.
That’s exactly what Willis, one of the biggest global names in the insurance market, recently put on the table.
The warning is straight to the point — companies are rolling out artificial intelligence solutions at scale, but the structures that should guide, control, and ensure responsible use of these technologies are still in their infancy. 🚨
This gap between what the technology can already do and what governance still doesn’t cover represents a real risk, both for insurers and for the consumers who depend on those automated decisions.
In this article, we take a deep dive into this landscape, break down what Willis identified, why it matters, and what’s at stake when innovation moves faster than the rules can keep up. 👇
What Willis Identified and Why It Turned Heads
Willis isn’t just any company in the industry. With decades of experience in the global insurance market and a broad perspective on corporate risk, when they raise a flag about AI governance, the market pays attention — and rightfully so.
The company’s recent positions point to a concerning scenario: while artificial intelligence tools are becoming increasingly sophisticated and embedded in underwriting processes, claims analysis, pricing, and customer service, the frameworks that should ensure all of this operates ethically, transparently, and in an auditable way are still far from covering the necessary ground. This isn’t a criticism of the technology itself, but rather of the unbalanced pace of this transformation.
The central point raised by Willis is that the speed of AI adoption in the insurance sector is creating a kind of gray zone, where important decisions — ones that directly affect people’s lives — are being made by algorithms that still lack proper oversight.
Picture an automated system denying health coverage or flagging a customer as high risk based on patterns that no human can clearly explain. This is already happening, and the absence of robust governance frameworks means there are no clear mechanisms to challenge, review, or correct these decisions efficiently.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Another relevant aspect that Willis brought to the conversation is the question of accountability. When an AI model makes a mistake that results in financial loss or consumer harm, who answers for it? The insurer? The technology vendor? The model developer? This lack of clarity around responsibility and accountability is exactly the kind of gap that well-structured governance frameworks should address, but in practice, it remains unanswered in most organizations across the sector.
Why Governance Is Falling Behind
Understanding why governance frameworks are lagging behind AI adoption requires looking at the internal dynamics of companies and the external regulatory environment at the same time.
On the internal side, there’s enormous pressure for efficiency, cost reduction, and competitive advantage. AI tools deliver measurable, fast results — whether it’s speeding up policy approvals, detecting fraud, or personalizing insurance products. Faced with these tangible gains, the natural tendency for organizations is to accelerate implementation and leave the discussions about control and oversight for later. The problem is that later often never comes, or it comes too late.
On the external side, the regulatory landscape is still under construction in virtually every major market. Regulators around the world — including in Brazil, with SUSEP, and in Europe, with the AI Act — are working to create specific guidelines for the use of AI in sensitive sectors like insurance. But regulation takes time, going through public consultation, review, and adaptation processes, while technology advances in cycles of months. This speed asymmetry creates exactly the vacuum that Willis is flagging: companies adopting AI on a massive scale before any legal or regulatory structure is ready to properly frame it.
There’s also a third factor that complicates this picture even further: the technical complexity of modern AI models. Unlike traditional software systems, where the decision logic can be audited line by line, machine learning models and deep neural networks operate in ways that even their own creators can’t fully explain.
This makes developing effective governance frameworks an enormous technical challenge that goes well beyond simply creating internal policies or documenting processes. It requires specialized technical expertise, interpretability tools, and an organizational culture oriented toward transparency — and all of that demands investment and time that companies aren’t always willing to commit before reaping the technology’s rewards. 🤖
What’s at Stake for the Insurance Industry
The insurance market has a characteristic that makes it particularly sensitive to AI governance failures: it deals directly with people’s most vulnerable moments. A health claim, a car accident, the loss of an asset — these are situations where the consumer is already in a fragile state and depends on a fair, transparent, and well-grounded response from the insurer.
When that response is generated by an algorithm without proper oversight, without explainability, and without clear dispute mechanisms, the potential for harm goes far beyond financial. It erodes trust in the industry as a whole, and trust, in the insurance market, is literally the product being sold.
Beyond the direct impact on consumers, the absence of solid frameworks also exposes insurers themselves to significant risks:
- Regulatory risks, with fines and sanctions as AI regulations gain traction around the world
- Reputational risks, when controversial automated decisions hit the media or result in lawsuits
- Operational risks, when models trained on historical data start reproducing problematic patterns at industrial scale
Willis, as a risk management specialist, understands better than anyone that ignoring these warning signs in the short term can generate enormous liabilities in the medium and long term.
It’s important to make clear that the alert from Willis is not a manifesto against AI adoption. Quite the opposite. Artificial intelligence has the potential to make the insurance industry more efficient, more accessible, and more fair — as long as it’s implemented the right way.
What’s being signaled is the urgent need to balance adoption speed with governance maturity, ensuring that organizations don’t only harvest the immediate benefits of technology while pushing its risks into the future. That balance is exactly what separates a successful digital transformation from a crisis waiting to happen. 💡
Governance Frameworks: What’s Missing and What Can Be Done
When we talk about governance frameworks for AI, we’re not talking about bureaucracy or stalling innovation. We’re talking about practical structures that define how models are developed, tested, monitored, and updated over time.
This includes fundamental elements like:
- Clear policies on which data can be used to train models
- Mechanisms to ensure that data is representative and free from problematic biases
- Performance metrics monitored on an ongoing basis
- Defining who has the authority — and the responsibility — to intervene when something goes off track
In the insurance industry, where AI decisions have a direct impact on people’s financial lives, these elements are not optional. They’re essential.
Some more mature organizations are already moving in this direction. Major international insurance groups have started creating internal committees dedicated to AI ethics, hiring model interpretability specialists, and adopting frameworks like the NIST AI Risk Management Framework or IEEE guidelines for autonomous systems.
In Brazil, initiatives like SUSEP’s recommendations on the use of predictive models and the ongoing discussions around national AI regulation are paving the way, but there’s still a considerable gap between what’s being discussed in technical forums and what’s actually being implemented in the day-to-day operations of insurers.
The Importance of Transparency and Explainability
One of the most critical pillars of any AI governance framework is the ability to explain how a model arrived at a particular decision. In the technical world, this is known as explainability, and in the insurance sector, this issue takes on an even more relevant dimension.
When an insurer denies coverage, adjusts a premium, or classifies a risk profile, the consumer has the right to understand the criteria behind that decision. Traditional actuarial analysis models, however complex they were, had logic that could be traced and explained. With increasingly complex deep learning models and neural networks, that traceability becomes a significant technical challenge.
Willis understands that without transparency, the industry runs the risk of creating a relationship of distrust with its customers. Nobody wants to receive an automated response without knowing the reason behind it, especially during stressful moments like filing a claim. Building explainability mechanisms isn’t just a best practice — in many jurisdictions, it’s becoming a legal requirement.
In Europe, for example, the AI Act classifies AI systems used in insurance as potentially high risk, which means there will be rigorous requirements for documentation, auditability, and explainability. Companies that get ahead of these requirements will be in a much more comfortable position than those that have to scramble to comply after the regulation takes effect. 📋
The Role of Organizational Culture in AI Governance
Having a perfect framework on paper is pointless if the company’s culture doesn’t support its application in practice. This is another point that deserves attention when we talk about the alert from Willis.
Effective AI governance depends on people who understand the risks, who feel responsible for automated decisions, and who have clear channels to report problems when they spot them. This requires training, constant internal communication, and above all, leadership that is committed to the topic.
At many insurers, AI adoption is still treated as a purely technology project, delegated to data or IT teams. When governance stays confined to those departments, it loses the holistic perspective needed to cover all the impacts an AI model can generate — from legal and regulatory issues to implications for customer experience and brand reputation.
The companies that are managing to move forward in a more balanced way are the ones treating AI governance as a multidisciplinary responsibility, involving areas like compliance, legal, operations, customer service, and even communications. This integrated approach allows risks to be mapped more comprehensively and responses to potential issues to be more agile and coordinated.
The Global Landscape and Lessons for the U.S. Market
The warning from Willis echoes a trend being observed in markets around the world. In the United States, state regulators have already begun requiring insurers to disclose when and how they use AI in their decision-making processes. In Europe, the AI Act is creating the most comprehensive regulatory framework on the planet for artificial intelligence systems, with specific requirements for high-risk sectors like financial services and insurance.
In the U.S., the regulatory landscape is still evolving at different speeds across state lines. The National Association of Insurance Commissioners, known as the NAIC, has been closely following international discussions and has already signaled its intent to incorporate specific guidelines for the use of AI models and predictive algorithms in the insurance market. At the federal level, conversations around AI regulation continue to advance, with the potential to directly impact how insurers can deploy these technologies.
For American companies, now is the time to prepare. Waiting for regulation to arrive before thinking about governance can be a risky strategy. The costs of retroactive adaptation — both financial and operational — tend to be significantly higher than the investments in governance structures built from the very start of the technology adoption process.
What the current situation demands is a mindset shift: treating AI governance not as a cost or an obstacle to innovation, but as a strategic component of the technology adoption itself. Companies that manage to build robust governance structures before scaling their AI solutions have a real competitive advantage — not just because they avoid risks, but because they build credibility with regulators, partners, and consumers.
In the insurance market, where regulation is an intrinsic part of the business, that credibility is worth far more than any short-term gain that a rushed implementation could deliver. 🎯
