ECB convenes emergency meeting on AI cyber threat to banks
The European Central Bank convoked an emergency meeting with European banks on 26 May 2026 to address cybersecurity risks posed by Claude Mythos, an AI model developed by Anthropic. The model is reportedly powerful enough to breach banking protection standards, with vulnerabilities potentially exploitable within 30 minutes of patch release.
What happened
The European Central Bank convoked an emergency meeting with European banks scheduled for 26 May 2026 to address cybersecurity risks posed by Claude Mythos, an AI model developed by Anthropic.
Frank Elderson, ECB board member and vice-president of the banking supervision council, explained that vulnerabilities in software patches released by major providers could be reverse-engineered within approximately 30 minutes, requiring banks to deploy patches much faster than current practice.
Reported impact
Claude Mythos is reportedly powerful enough to breach the protection standards currently used by European banking institutions.
The rapid speed at which patch vulnerabilities can be exploited poses a significant challenge to current bank cybersecurity incident response processes.
Index analysis
Anthropic has provided access to Mythos to a limited number of non-European organizations, primarily based in the United States, as part of a testing programme called 'Project Glasswing'.
Governance controls
The ECB intends to press banks to accelerate solutions and emphasized the necessity for U.S. banks with more advanced technologies to share information with European institutions.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.