Anthropic’s Mythos AI Might Threaten Global Financial Resilience

During the recent spring meetings of the International Monetary Fund (IMF) and the World Bank, several global financial leaders issued stark warnings, stating that Anthropic’s latest AI model, “Claude Mythos,” possesses unprecedented vulnerability detection capabilities and could pose a “systemic threat” to global financial infrastructure. However, the UK banking sector has already been granted permission to begin using this technology starting next week.

According to system cards released by Anthropic, “Claude Mythos” can identify system vulnerabilities that have existed for decades within seconds and can even autonomously simulate complex cross-operating-system attacks. Anthropic’s CEO has publicly acknowledged that the model possesses capabilities “sufficient to surpass most top human hackers,” leading the company to decide against making it publicly available for now, limiting its use to defensive testing by a select few partners.

Canadian Finance Minister François-Philippe Champagne compared the threats posed by AI to traditional geopolitical crises. He noted: “The crisis in the Strait of Hormuz is a risk we can see and touch; but Anthropic’s technology represents ‘unknown unknowns.’” He emphasized that finance ministers worldwide must take this matter seriously and establish appropriate regulatory safeguards to ensure the resilience of the global financial system remains unshaken.

Bank of England Governor Andrew Bailey also stated that this poses a severe challenge for regulators. He warned that the pace of development in the AI field is astonishing; if regulation is implemented too early, it may hinder technological progress; if too late, it can lead to a loss of control over the situation.