IMF raises alarm on finance sector risks after Claude Mythos, wants international cooperation

The International Monetary Fund (IMF) issued a warning in May 2026 about systemic risks to global finance from advanced AI models like Anthropic’s Claude Mythos. The AI, announced in April 2026, was designed to identify cybersecurity flaws across software and operating systems but was not released publicly due to its perceived danger. The IMF stated that such AI models could dramatically reduce the time and cost for bad actors to exploit vulnerabilities, increasing the risk of coordinated attacks on financial systems. Claude Mythos can detect thousands of cybersecurity flaws that humans miss, raising concerns that attackers could exploit similar weaknesses across institutions using shared tools. The IMF warned this could lead to ‘correlated failures,’ disrupting payments, accounts, and financial operations globally. Currently, Mythos is restricted to around 40 companies under Anthropic’s Project Glasswing. The IMF highlighted that AI-driven attacks operate at machine speed, outpacing patching and remediation efforts. It also noted that advanced AI may inevitably breach existing cyberdefenses, putting sectors like energy, telecom, and public services at risk. Emerging and developing economies face higher threats due to limited resources, potentially allowing attackers to weaken global systems. To mitigate risks, the IMF called for international cooperation, including AI-driven safeguards and information sharing. Governments like the US and India have already expressed concerns, with the US considering an AI oversight committee to review models before launch. The IMF emphasized that global collaboration is essential to prevent widespread disruptions from AI-driven cyber threats.