Responsible AI in insurance: Trust, accountability and resilience in a changing landscape

Early AI adoption in the insurance industry focused on efficiency, automation and cost reduction. But as AI becomes more deeply embedded in decision-making, from claims processing to customer engagement, the conversation is shifting. Speed alone is no longer the differentiator. Trust is.

In this episode of From Transactions to Trust, Thomas Rauschen and Tom Infante explore how insurers can design, deploy and scale AI responsibly, balancing innovation with ethics, governance and resilience.

Moving beyond efficiency: Why trust defines AI success

AI adoption across insurance is accelerating, driven by the need for digital transformation and modernization. At the same time, organizations face increasing pressure to strengthen cybersecurity, resilience and regulatory compliance.

But as AI systems begin influencing outcomes at scale, the stakes change. Responsible AI is about building powerful models and ensuring those systems are safe, aligned with human values and designed with the end user in mind.

This shift mirrors a broader trend seen across industries: success with AI is no longer measured purely by efficiency gains, but by how well organizations ensure accountability, governance and trust in AI-driven decisions.

From security to resilience: A new operational imperative

Historically, security and resilience were treated as separate disciplines. Security focused on prevention, while resilience focused on recovery. Today, that distinction is disappearing.

Organizations must be both secure and continuously resilient—capable not only of protecting systems, but also of adapting, responding and recovering in real time. This is particularly critical in AI-enabled ecosystems, where interconnected systems amplify both opportunity and risk.

Defining responsible AI in practice

Responsible AI begins with a simple principle: systems must be designed to serve people.

In practice, this means:

• aligning AI models with human values
• ensuring fairness and mitigating bias
• maintaining transparency and explainability
• embedding accountability across the value chain

One of the biggest challenges is ownership. In complex ecosystems involving multiple vendors and models, accountability can become unclear. Yet responsibility ultimately sits with those who design and deploy the AI capability.

Maintaining a “human in the loop” remains essential, not as a bottleneck, but as a safeguard to continuously assess outcomes and ensure ethical alignment.

As Infante emphasizes, “we should design and build systems and models that are not only safe… but aligned with human values.”

Governance as the foundation of trust

As AI adoption scales, governance becomes a central concern.

Key risks include:

• Automation bias: unintentionally reinforcing historical or data-driven biases
• Ethical drift: gradual deviation from organizational values as systems evolve
• Lack of transparency: limited visibility into how decisions are made

To address these challenges, organizations must embed governance into AI design from the outset. This includes clear oversight mechanisms, transparent processes and continuous validation of outcomes.

Importantly, regulation alone is not enough. Given the speed of technological advancement, insurers cannot rely solely on external regulators. They must build internal frameworks that proactively manage risk, ensure compliance and protect customer trust.

In this episode of CGI's "From Transactions to Trust," we move beyond the hype of AI in banking to discuss what really matters: accountability and revenue. While AI has long been seen as a tool for efficiency, the rise of "agentic AI" is changing the game. These intelligent, semi-autonomous systems are not just optimizing processes—they are making decisions, generating revenue, and forcing a conversation about accountability.

Join our experts, Frederic Miskawi (Vice-President, Global Applied AI Lead) and Gaby Martin (Director, Consulting Services – U.S. Operations), as they outline a practical roadmap for building an "agentic bank." You'll learn how to:

• Generate real, measurable ROI from your AI initiatives.
• Transform traditional cost centers into engines of revenue.
• Accelerate the modernization of your legacy systems.
• Navigate the shift from rule-based automation to dynamic, intelligent action.

This episode is a must-listen for banking leaders, technologists, and anyone interested in the future of financial services.

We unpack why AI value in insurance stalls when organisations treat legacy as a technology problem instead of a foundation problem across data, process and mindset. We map a practical path from quick efficiency wins to intelligence-driven modernisation that actually changes decisions and operating models.
• legacy as systems plus operating model plus mindset
• avoiding the trap of layering AI over old processes
• efficiency-driven AI wins in underwriting document handling
• moving from faster work to better decisions with AI support
• AI ambition running ahead of data readiness and modernisation reality
• iterative test-and-learn guided by a north star
• starting with the business problem then assessing maturity
• portfolio approach to pick easy proofs and harder high-value bets
• AI as a consideration alongside operational resilience and cyber security
• success factors: clarity of purpose, human change, constant relevance checks