BoE and ECB warn AI could threaten financial stability and cyber resilience

Europe’s top central bankers are now naming AI as a distinct source of systemic risk, not just a background technology trend. The BoE is worried both about an AI-fueled asset bubble and about cyber vulnerabilities as banks plug more models into core systems. The ECB is going further, instructing lenders to deliver concrete action plans against AI-enabled cyberattacks within four months. Together, these moves signal that highly capable models like Mythos are being treated as potential offensive tools that could attack payment rails and erode confidence in banks.

For the AGI race, the message is that financial regulators will not wait for clear failure modes before intervening. As models gain agentic capabilities, the bar for acceptable operational and cyber risk in critical infrastructure will climb, and AI vendors will face heavier due diligence from banks and supervisors. That may slow the deployment of frontier models into regulated finance, but it also strengthens the argument that only a handful of very well-capitalized labs can realistically meet the security expectations that are emerging.