The central bank that regulates trillions in assets can't get its hands on the AI tools its banks need to defend themselves.
The Summary
- Bank of England Governor Andrew Bailey confirmed British banks remain locked out of Anthropic's Mythos AI tool, speaking at a central banking conference in Iceland
- Bailey called for a multinational approach to cybersecurity threats, acknowledging UK banks are using other AI models to test their cyberdefenses
- The access gap reveals a sovereignty problem: critical financial infrastructure depends on AI tools it can't control or even access
The Signal
Bank of England Governor Andrew Bailey told Bloomberg's Stephanie Flanders that British banks still can't use Mythos, Anthropic's latest AI tool built for complex threat modeling and cybersecurity defense. This isn't a temporary delay or a pricing negotiation. It's a formal lockout of an entire national banking system from what appears to be a tier-one defense capability.
The timing matters. Bailey made these comments at a central banking conference in Iceland, where cybersecurity sits high on the agenda. Central bankers don't usually air their AI procurement problems in public. When they do, it's because the workaround isn't working.
"British banks are still locked out of using Anthropic's powerful new AI tool."
Bailey noted that UK banks are using other models to test their cyberdefenses, which sounds fine until you consider what that means. If Mythos represents a step-function improvement in threat detection or response simulation, then "other models" translates to "second-tier defenses." In cybersecurity, second-tier means you're defending against yesterday's attacks with yesterday's intelligence.
The governor's response is telling: he's calling for a multinational approach to fighting cybersecurity threats. That's central banker code for "we can't solve this alone, and the current setup isn't working." When a G7 central bank governor says the system needs restructuring, what he means is that national regulators no longer control the tools their regulated entities need to function safely.
The Implication
This is what AI sovereignty looks like when you don't have it. The Bank of England oversees institutions holding trillions in deposits and facilitating the majority of UK economic activity. Those institutions need cutting-edge AI to defend against adversaries who definitely have cutting-edge AI. And they're stuck waiting for access while a private company in San Francisco decides who gets what and when.
Watch for two things. First, whether other central banks face similar lockouts or if this is UK-specific. Second, whether Bailey's call for a multinational approach leads to actual frameworks or just more conferences. The gap between what regulators need and what they can access is widening. Someone will close it, either through international coordination or through domestic alternatives that probably won't be as good.