After the US, UK Financial Regulators Will Discuss Anthropic's New Model Risk with Banks
Anthropic's latest AI model, Mythos, sparks a global financial regulatory storm: the Federal Reserve joins forces with the Treasury Secretary to urgently convene Wall Street bank CEOs; the Bank of England and Bank of Canada follow suit, marking rare coordination among the three nations' regulators within days. This high-complexity model, capable of automatically identifying and exploiting vulnerabilities in mainstream operating systems, has been listed by all parties as one of the most significant cybersecurity threats facing the financial industry today
The regulatory alarms triggered by Anthropic's latest AI model Mythos are spreading across the Atlantic. Following US Treasury Secretary Bessent and Federal Reserve Chair Powell's urgent convening of Wall Street bank CEOs, the Bank of England has also added this model to its agenda, preparing to formally discuss its potential risks with financial institutions.
According to Bloomberg's report last Saturday, the Bank of England plans to specifically discuss the impact of the Anthropic Mythos model on financial institutions within the next two weeks at its Cross-Market Operations Resilience Group (CMORG) and CMORG Artificial Intelligence Working Group meetings. Meetings will invite representatives from HM Treasury, the Financial Conduct Authority (FCA), and the National Cyber Security Centre (NCSC) to participate.
This series of regulatory actions reflects the shared judgment of regulators in various countries: cyber attack threats catalyzed by new AI models represented by Mythos have become one of the most significant risks currently faced by the financial industry.
US-UK Regulatory Coordination, Alert Spans Across the Atlantic
According to Bloomberg's report last Friday, US Treasury Secretary Bessent and Federal Reserve Chair Powell convened CEOs of major Wall Street banks at the Treasury Headquarters in Washington on Tuesday this week, where the core agenda was precisely the cybersecurity threats potentially posed by Mythos. According to informed sources, the purpose of this meeting was to ensure each bank fully understood the potential risks of Mythos and similar models, and had taken necessary system protection measures.
This temporary arrangement for the US meeting, without prior public reporting, itself sends a strong signal. All invited banks are listed by top regulators as Systemically Important Financial Institutions, whose stability directly concerns the global financial system.
Britain followed suit immediately. According to an informed source, relevant meetings of the Bank of England have already been scheduled. The Bank of Canada also held meetings with banks and financial institutions last Friday, specifically discussing the cybersecurity risks brought by Mythos.
Mythos Model: Capabilities and Risks Coexist
Anthropic stated that Mythos is a highly complex AI model capable of identifying and exploiting vulnerabilities in all mainstream operating systems and network browsers. It is precisely this capability that has triggered high vigilance among financial regulatory layers.
The Trump Administration issued warnings to bank executives this week, requiring them to take the model seriously and deploy relevant capabilities to detect vulnerabilities in their own systems. Currently, major Wall Street banks have begun internal testing of Mythos.
The message conveyed by the above series of emergency meetings is highly consistent: Regulators believe that new AI-driven cyber attacks have risen to become one of the biggest risks currently faced by the financial industry. From the US to the UK to Canada, financial regulators of major economies acted sequentially in a short time frame; this collaborative posture is not common in the history of financial regulation in recent years.
