European Central Bank supervisors will question bankers regarding the risks posed by Anthropic's new artificial intelligence model, Mythos. Cybersecurity experts believe the model could significantly challenge the banking industry's legacy technology systems, triggering alarm among regulators in Britain and the U.S.
ECB supervisors are collecting information to assess banks' preparedness for this emerging risk. While the U.S. has seen urgent meetings with bank executives and warnings from Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell, the ECB's approach will involve regular dialogue with bank staff.
Experts note Mythos' advanced coding capabilities could enable it to identify and exploit cybersecurity vulnerabilities at an unprecedented speed and scale. Anthropic has stated the current iteration, Claude Mythos Preview, will not be generally available. Instead, Project Glasswing invites major tech companies and financial institutions like JPMorgan Chase to privately evaluate the model and develop defenses.
U.S. officials, including President Donald Trump, have backed government safeguards, and St. Louis Fed President Alberto Musalem emphasized the need for the central bank to re-evaluate its approach to cybersecurity. Britain's Technology Secretary and Security Minister also issued a warning, stating Mythos is "substantially more capable at cyber offence" than previously tested models.
Central banks globally are racing to understand the implications. Bank of England Governor Andrew Bailey stressed the urgency, and the model was discussed at a recent Canadian finance meeting involving ministry and Bank of Canada representatives.