European Central Bank supervisors are preparing to question bankers regarding the risks associated with Anthropic's new artificial intelligence model, potentially supercharging cyberattacks. The model, known as Mythos, is raising alarms among regulators in Europe, Britain, and the United States due to its advanced coding capabilities. Cybersecurity experts believe Mythos could identify and exploit vulnerabilities in banking systems at an unprecedented speed and scale.
ECB supervisors are collecting information to assess banks' readiness for this emerging threat. Anthropic has stated that the current iteration, Claude Mythos Preview, will not be released publicly. Instead, the company launched Project Glasswing, inviting major tech firms, cybersecurity vendors, and institutions like JPMorgan Chase to privately test the model and develop defenses.
U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell recently met with bank executives to discuss these risks. President Donald Trump has acknowledged the threats and supports government safeguards. In Britain, Technology Secretary Liz Kendall and Security Minister Dan Jarvis issued similar warnings, noting Mythos is significantly more capable in cyber offense than previously tested models. Bank of England Governor Andrew Bailey emphasized the urgent need for central banks and regulators to grasp the implications of such advanced AI. Tech risk is already a top priority for the ECB.