Anthropic will brief the Financial Stability Board on serious cyber weaknesses its new Claude Mythos model has found in global financial systems. The FSB unites top finance ministries, regulators, and central banks from G20 economies and other major jurisdictions.
The watchdog requested the briefing after Bank of England governor Andrew Bailey, who chairs the FSB, warned that Mythos could pose “major security risks” to a banking sector still running on aging technology. Regulators now want Anthropic to explain what the model can do and where it sees the most dangerous flaws.
Claude Mythos is a cybersecurity-focused version of Anthropic’s Claude AI that scans software, operating systems, and infrastructure for vulnerabilities. Anthropic says Mythos Preview has already found thousands of high‑severity bugs, including issues in every major operating system and web browser.
Some of the vulnerabilities it uncovered are as old as 27 years and had never been flagged by developers or watchdogs. Cyber experts say this shows how frontier AI can now outperform most humans at spotting and even exploiting flaws in code, especially in complex, older systems that banks still depend on.
Why Mythos Worries Financial Regulators
Governments started paying close attention after reports that Mythos can help carry out multi‑stage cyberattacks in controlled lab tests. The UK’s AI Safety Institute found that Mythos Preview could autonomously discover and exploit weaknesses on small, poorly defended networks, completing advanced capture‑the‑flag challenges that would take human professionals days.
At the same time, Anthropic is investigating claims that a small group gained unauthorized access to Mythos through a third‑party vendor, even though there is no sign that criminals hold the model. The company has already limited access to about 40 organizations, including major tech and security firms, through a program called Project Glasswing, rather than releasing Mythos broadly.
Finance ministers and central bank chiefs want Anthropic to map out how Mythos sees today’s attack surface for banks and payment networks. They also want clear guidance on how to harden aging core systems before similar models spread more widely and fall into hostile hands.
US and Indian officials have already summoned bank leaders to emergency meetings and urged them to test critical systems against Mythos‑class threats. With the FSB now involved, G20 regulators see Claude Mythos as both a powerful defensive tool and a potential offensive weapon, and they expect Anthropic to show how to manage that risk without slowing vital AI security research.
READ MORE: Hyperliquid Price Jumps 10% After HYPE ETF Debut: $54 Target Looms
