AI Risk in Banking: What Jamie Dimon’s Initiative Cannot Fix
CCO Managing Partner
Jamie Dimon is personally recruiting bank and technology leaders into an expanded industry group to tackle AI risks. The initiative grows out of the Alliance for Critical Infrastructure and already reaches more than 40 companies across banking, energy, utilities, telecom, airlines and other critical sectors.
The trigger is clear. Advanced models are raising real concerns about cyber exposure, systemic vulnerability and uncontrolled capability. Dimon’s own line about “giving ballistic missiles to individuals” captured the mood. When the CEO of the largest bank in the United States treats AI risk as a collective priority, the rest of the industry pays attention.
That focus is necessary. Shared understanding of model risk, coordinated safeguards and better dialogue with regulators all make sense. External and systemic risks are real and growing.
But they are not the only risks that matter.
The Quieter AI Bottleneck Inside Banks
Inside most banks, the more immediate constraint looks different.
AI is already accelerating analysis, scenario work and reporting. What often remains unchanged is the time from insight to decision. Better information arrives faster, then still waits in the same alignment meetings, approval chains and unclear ownership structures.
At the same time, a large part of institutional knowledge still lives in people’s heads and in undocumented local logic. Definitions of key metrics differ across teams. Business rules sit in Excel files and tribal knowledge. When agents start operating on that foundation, they inherit the ambiguity.
The result is familiar: impressive pilots, growing use-case inventories and limited impact on actual decision speed or decision quality.
What Industry AI Risk Groups Cannot Fix
Cross-industry coordination can improve how institutions share threat intelligence and align on safeguards. It cannot redesign decision rights inside a single bank.
It cannot force clarity on what “revenue”, “risk” or “client value” actually mean across systems.
And it cannot create ownership of the knowledge layer that both humans and agents need to work reliably.
These are internal operating model questions. They require leadership that is willing to treat AI as more than a technology or risk topic.
They require deliberate choices about:
- Who owns the definitions and the business rules
- Which decisions remain human and why
- How exceptions are escalated and documented
- Whether the organisation is prepared to act at the new speed that AI enables
Without those choices, even the best external AI risk framework leaves the internal foundation unchanged.
The Real Test of AI Readiness in Banking
Dimon’s initiative is a signal that AI risk has moved from the innovation lab to the executive agenda.
That is progress.
The next test is whether the same seriousness is applied inside the institution. Not only to model risk and cyber, but to the quieter questions of meaning, ownership and decision rights.
External AI risk is getting organised.
Internal AI readiness is still the quieter bottleneck.
The banks that close both gaps will be the ones that turn AI from an expensive experiment into a durable advantage.
If you are asking where the internal bottlenecks sit in your own organisation, book a 30-minute exploration call with us. We can look at your current AI operating model, decision structures and knowledge foundation, and identify where stronger internal readiness could unlock more value from the AI initiatives already underway.