The shockwaves from Claude Mythos Preview have crossed the Atlantic. After US regulators at the Fed and Treasury responded last week, British financial authorities are now stepping in. And they’re taking it seriously.
Bank of England gets involved
According to Bloomberg and the Financial Times, the Bank of England, the Financial Conduct Authority (FCA), and the UK Treasury are planning joint discussions with the National Cyber Security Centre (NCSC). The goal: assess the risks that Claude Mythos Preview poses to the British financial system.
The discussions will take place within the Cross-Market Operational Resilience Group (CMORG). This isn’t a routine channel. CMORG was created for systemic risks that could affect the entire financial market.
What’s worrying regulators
The core issue: Anthropic has publicly stated that Mythos Preview has identified thousands of severe security vulnerabilities in operating systems, browsers, and widely-used software. Many of these were previously unknown. That’s precisely why Anthropic is only making the model available through Project Glasswing in a restricted fashion. But it also means the discovered vulnerabilities exist regardless of whether Mythos is publicly available or not.
Representatives from major British banks, insurers, and exchanges are expected to be briefed on the cyber risks within the next two weeks.
A globally coordinated response
What we’re seeing is an unprecedented coordinated response from global financial regulators to an AI model. In the US, Treasury Secretary Bessent and Fed Chair Powell have already invited bank CEOs for a briefing. Now the UK follows. Regulators are increasingly treating the capabilities of advanced AI models as a potential systemic risk.
My take
This is new territory. We’ve seen regulators discuss AI regulation before. But a central bank calling an emergency meeting because of a specific AI model is without precedent. It shows how seriously Mythos’s capabilities are being assessed.
For Anthropic, it’s a double-edged sword. On one hand, the response proves that Project Glasswing was the right call. On the other, regulatory pressure is mounting. The question of whether and when Mythos-class models will be released for general use won’t be answered by Anthropic alone. The regulators are at the table now.
Sources: