Anthropic to Brief FSB on Mythos AI Vulnerabilities After White House Restricts Distribution

Anthropic will brief the Financial Stability Board on critical cybersecurity vulnerabilities uncovered by its new Mythos Preview AI model, following a request from Bank of England Governor Andrew Bailey, who chairs the FSB. Bailey warned that the model could potentially crack the whole cyber risk world open. Mythos is designed to detect long-standing vulnerabilities in software, infrastructure, and web browsers, and Anthropic claims it has found thousands of high-severity flaws across every major operating system and web browser, posing a severe threat to global banking infrastructure that often relies on legacy technology.

Anthropic has released the model to approximately 40 organizations, including Microsoft, Amazon, JPMorgan Chase, and Mozilla. Mozilla alone used Mythos to identify and patch 423 Firefox bugs. However, further distribution has been halted at the request of the Trump administration, which asked Anthropic to limit wider release due to security risks. The FSB is currently developing a report on sound practices for AI adoption in the financial system, while the International Monetary Fund has warned that such AI models could elevate cyber risk to a potential macro-financial shock. Financial experts and regulators warn that if adversaries or state-sponsored actors obtain similar AI capabilities, they could weaponize these vulnerabilities to destabilize global financial stability.