US Officials Warn Banks of AI Model Mythos Cybersecurity Risks
US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with leaders of major banks to warn them about cybersecurity risks from Anthropic's new AI model Mythos, which reportedly has capabilities to identify and exploit vulnerabilities in operating systems and web browsers.
What happened
Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell called an emergency meeting in Washington with leaders of major US banks including Citigroup, Morgan Stanley, Bank of America, Wells Fargo, and Goldman Sachs to warn them about cybersecurity risks from Anthropic's new AI model Mythos.
According to US authorities, the Mythos model reportedly has the capability to identify and exploit vulnerabilities in major operating systems and web browsers, potentially enabling a new era of cyberattacks. Anthropic confirmed the model has both offensive and defensive capabilities in cyberspace and has restricted initial access to approximately 40 large technology and financial companies.
Reported impact
US authorities fear the Mythos system could open the way to a new era of cyberattacks. During testing, the system reportedly found ways to access data from different browsers and identified potential security gaps that could theoretically affect banking websites.
Index analysis
Regulatory authorities view AI-enabled cybersecurity threats as among the largest risks to the global financial system. The shift from traditional cyberattacks to advanced, AI-guided tools that can identify vulnerabilities means banking defense cannot rely solely on existing security practices but requires continuous adaptation.
Governance controls
Anthropic chose not to pursue broad distribution of the model and restricted access to approximately 40 technology companies. The company is in dialogue with US government bodies about the system's capabilities and has informed officials and key market stakeholders in advance.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.