British banks are currently encountering significant obstacles in gaining access to Anthropic PBC’s cutting-edge Mythos artificial intelligence model, a platform designed to enhance the detection and defense against cyber threats. Bank of England Governor Andrew Bailey recently brought attention to this issue during an interview, expressing concern over the delays and emphasizing the vital role such AI technology plays in protecting financial systems.
Bailey revealed that although Anthropic has offered a trial sharing of the Mythos AI model, political and regulatory barriers—particularly involving U.S. government policies—have prevented UK banks from fully utilizing it. The exact causes of these delays remain somewhat unclear but appear connected to contrasting national approaches to AI regulation and cybersecurity oversight.
Speaking at a central banking conference in Reykjavik, Bailey stressed the importance of international cooperation in addressing cyber risks. He warned that cyber threats do not respect borders and cannot be effectively managed through isolated national efforts. “The spillovers from this type of cyber risk are so significant that relying on a single country’s approach is insufficient,” he said.
The Mythos AI model has attracted widespread attention for its potential to transform how vulnerabilities within financial infrastructures are detected and managed. Last month, Bailey suggested that Anthropic might have “discovered a way to unlock new dimensions in cyber risk management.” Despite this optimism, some cybersecurity experts caution that fears about the model enabling unrestricted hacking capabilities may be exaggerated.
Compounding the issue is the recent delay by U.S. authorities of an executive order intended to create a voluntary framework for AI developers to coordinate with federal regulators before releasing advanced models publicly. Anthropic is also involved in ongoing discussions with U.S. officials regarding guidelines for military applications of its AI technologies, adding further complexity to an already challenging regulatory environment.
Meanwhile, UK financial institutions continue to deploy alternative AI tools to test and strengthen their cybersecurity defenses but remain unable to access Mythos. Bailey, who also chairs the Financial Stability Board internationally, reiterated that confronting cybersecurity challenges requires coordinated global standards and collective action rather than fragmented national policies.
This ongoing delay underscores broader tensions between rapid technological innovation, evolving regulatory frameworks, and geopolitical factors as governments and financial institutions navigate the complex landscape of artificial intelligence and cyber risk management.
