The news
A recent episode of Bloomberg Tech: Europe, recorded at the Wave by Vento tech conference in Turin, centers on artificial intelligence adoption inside banks. The program examines both the operational gains banks seek and the new vulnerabilities that surface when models handle higher-stakes choices. Guests include JPMorgan CEO Jamie Dimon, Revolut CEO Nik Storonsky, Evident CEO Alexandra Mousavizadeh, and Bending Spoons CEO Luca Ferrari.
The discussion frames AI as a tool already shifting day-to-day banking processes toward greater speed and lower cost. At the same time, participants note that wider use of the technology creates exposure points not present in earlier rule-based systems. The episode runs as a monthly, 30-minute magazine-style broadcast focused on European technology leaders, investors, and policy questions amid rising global competition.
Context
Prior coverage of banking technology tended to treat AI as an experimental add-on for fraud detection or customer chat. The Turin program marks a shift toward treating the technology as a core operating layer that influences credit decisions, compliance checks, and internal risk models. European banks now face simultaneous pressure from U.S. and Asian competitors who have moved faster on model deployment.
The change affects both large incumbents and newer digital banks. Traditional institutions must retrofit existing compliance frameworks, while fintechs such as Revolut already run many customer-facing processes through live models. Regulators and security firms track the same rollout for signs that concentrated model failures could produce systemic effects.
Details
Tom Mackenzie hosts the segment and structures the conversation around concrete use cases now moving from pilot to production. Dimon addresses JPMorgan’s internal deployment patterns, while Storonsky describes how Revolut scales models across payments and onboarding. Mousavizadeh brings data from Evident on how third-party AI vendors are evaluated for resilience, and Ferrari outlines engineering choices at Bending Spoons that affect downstream financial services.
The program explicitly links efficiency claims to risk questions. Faster transaction monitoring and automated credit scoring reduce headcount and latency, yet the same automation concentrates decision logic inside opaque models. When those models err, the blast radius can reach regulatory capital, customer funds, or market stability. No numerical benchmarks are supplied in the episode, but the guests treat the tradeoff as the central unresolved tension.
The broadcast also situates the topic inside broader European tech competition. The show positions the Turin conference as one node in a wider effort to keep pace with developments outside the region. Interviews with the four executives therefore serve both as case studies and as signals of where capital and talent are flowing.
Why it matters
For software engineers and technical founders working in or adjacent to financial services, the episode underscores that model deployment is no longer a research exercise. Production systems now carry direct balance-sheet and regulatory weight, which changes requirements for testing, monitoring, and rollback. Teams that previously optimized only for accuracy must now design for auditability and containment when models encounter edge cases.
The discussion also signals that security and compliance roles will expand rather than shrink. New vulnerabilities are not primarily about external hacks but about internal model drift, data lineage failures, and correlated errors across institutions that rely on similar foundation models. Organizations that treat these issues as afterthoughts will face higher remediation costs once live traffic exposes them.
The program leaves the regulatory response open. European supervisors have so far issued principles rather than detailed technical standards. Until those standards crystallize, banks and their vendors operate under uneven expectations about explainability and stress testing. The practical result is that engineering roadmaps must carry extra margin for future compliance work that is not yet fully specified.
---
Sources:
No comments yet