British financial regulators launched emergency talks this week to assess cybersecurity risks from Anthropic’s unreleased Claude Mythos AI model, following similar urgent meetings in Washington days earlier. The Bank of England, Financial Conduct Authority and HM Treasury are coordinating with the National Cyber Security Centre and major UK banks to evaluate vulnerabilities exposed by the AI system.
Representatives from British banks, insurers and exchanges will receive formal briefings within the next fortnight on potential threats to critical financial infrastructure.
The coordinated response comes after US Treasury Secretary Scott Bessent summoned chief executives from Goldman Sachs, Bank of America, Citigroup, Morgan Stanley and Wells Fargo to Washington earlier this week. Federal Reserve Chair Jerome Powell attended that meeting focused on what regulators described as unprecedented cybersecurity risks posed by Anthropic’s latest model.
Anthropic’s Claude Mythos Preview has already identified thousands of major vulnerabilities across operating systems, web browsers and widely used software applications. Some of the security flaws discovered date back 27 years and had gone unnoticed by human security researchers until the AI system flagged them.
In response to these findings, Anthropic has taken the unprecedented step of restricting access to its new model under “Project Glasswing,” a controlled initiative allowing only select organizations to use the technology for defensive cybersecurity purposes. Amazon, Apple, Microsoft, Cisco, Broadcom and the Linux Foundation have gained early access under strict limitations.
This marks the first time Anthropic has imposed such restrictions on any of its products.
“AI models have now surpassed all but the most skilled humans at finding and exploiting software vulnerabilities,” the company stated in a blog post earlier this month, warning that “the fallout, for economies, public safety, and national security, could be severe.”
JP Morgan CEO Jamie Dimon echoed these concerns in his annual shareholder letter published this week, noting that cybersecurity “remains one of our biggest risks” and that “AI will almost surely make this risk worse.”
The UK discussions center on how vulnerabilities identified by Mythos could affect critical IT systems underpinning financial markets. Regulators are examining whether weaknesses in widely used software could be exploited to disrupt banking operations or compromise sensitive financial data.
While no immediate threat has been confirmed, the speed of regulatory response across multiple countries reflects growing concern about AI’s potential to reshape cybersecurity threats at scale. The Bank of England has placed the issue on its resilience taskforce agenda as authorities encourage financial institutions to adopt advanced tools against emerging AI-driven risks.
Leave a comment