AI in Finance Faces Tough Questions on Risk and Readiness

There’s a growing divide between ambition and reality when it comes to AI adoption strategies within risk management that reveals how cultural barriers, governance gaps, and poor data quality continue to limit its true impact on modern finance.

That was the takeaway from a panel discussion at Risk Evolve 2025, CeFPro’s flagship European conference event.

The conversation brought together perspectives from venture capital, investment banking, and operational risk, underlining just how differently institutions experience the promise and the pressure of AI. 

For some, AI is about survival in high-velocity markets. For others, it’s a tool to help make sense of unstructured information where traditional models fall short.

In venture capital, the challenge is clear. Without historical performance to lean on, assessing risk requires forward-looking judgment. 

Here, AI is less about replacing decision-makers and more about enhancing them - filtering vast information flows into usable insights. But even its champions admit the technology adds little when innovation itself is unpredictable and non-linear.

On trading floors, the story looks very different. Data is abundant and tools exist to parse it, from sentiment tagging to automated reporting. 

Yet the transformative vision of AI as an autonomous agent - capable of querying databases, performing calculations, and offering actionable decisions in real time - remains out of reach. 

Fragmented systems, patchy infrastructure, and mismatched models leave the dream unfulfilled.

Governance is proving just as knotty. Which use cases are worth pursuing? Who is accountable when AI fails? 

Pressure to deploy is mounting from regulators, boards, and competitors alike, but without an inventory of use cases or a shared language across business units, many deployments risk being cosmetic. 

Risk teams are increasingly called on not only to gatekeep but to educate, bridging the knowledge gap across departments.