Read, Watch or Listen

United States-based artificial intelligence company Anthropic is preparing to brief members of the Financial Stability Board on cyber vulnerabilities its latest frontier AI model, Mythos, identified within parts of the global financial system, according to a Financial Times report cited by Reuters on Monday. The discussions are expected to focus on weaknesses uncovered during testing of Mythos, a model that has drawn attention in the cybersecurity sector for its reported ability to identify complex vulnerabilities across multiple software environments. The Financial Stability Board, which brings together regulators, central banks, finance ministries and supervisory authorities from major economies, has not disclosed which institutions or systems are involved, and Reuters said it could not independently verify the Financial Times report. The briefing comes as regulators in the United States and Europe warn that concentrated cyber risks inside interconnected financial networks could trigger cascading disruption if exploited across multiple institutions. Washington, United States, is also a focal point for a broader debate over how rapidly improving AI systems are changing the cybersecurity landscape, with governments and security agencies increasingly concerned about models that can identify software flaws, network weaknesses and potential attack paths inside critical infrastructure. In recent weeks, technology firms and researchers have reported that frontier AI systems are becoming more capable of finding previously unknown vulnerabilities, generating exploit code and automating parts of cyber operations. Google’s threat intelligence division separately reported a case in which a cybercrime group used AI tools to help discover a new software flaw and build an exploit for a widely used open-source administration platform, an attack that was stopped before broad deployment. Against this backdrop, Anthropic’s planned briefing to the Financial Stability Board adds pressure on policymakers already assessing how to manage frontier AI risks in global finance, critical infrastructure and cybersecurity, even as major banks expand their own use of AI for trading, fraud detection and compliance.

By Daniel Hayes | JQJO News