The security layer in global financial infrastructure operates on an asymmetry: attack costs fall with automation while defense costs scale with the complexity of the systems being protected. That asymmetry is why Bank of England Governor Andrew Bailey's warning about frontier AI matters beyond the usual regulatory noise. Bailey has raised the alarm that new AI models could materially increase cyber risks to the global financial system.
Modern financial networks are deeply coupled. Payment rails, clearing houses, interbank messaging systems, and trading infrastructure share interconnects that span jurisdictions and institutions. A compromise that propagates through that stack fast enough to outrun incident response can become a systemic event, not a firm-level loss.
The threat that frontier AI introduces at this layer is primarily about speed and scale on the offensive side. Advanced models can accelerate attack planning and vulnerability scanning in ways that raise attacker throughput without proportionally raising attacker cost. That shifts the operational posture of every institution sitting in the stack.
Bailey's decision to frame this as a threat to global financial stability, rather than as a firm-level operational risk, points to his concern about the interconnect. Single-institution cyber incidents, even large ones, are containable. Attacks that move across the network faster than coordination can happen are not.
The Bank of England has not specified whether the governor's warning will translate into new supervisory requirements or multilateral policy coordination. What Bailey has done is place the question at the level of systemic risk, which is the correct altitude for a constraint this structural.