Canada regulator warns banks of Claude Mythos cyber risk

Canada's top financial regulator has warned the nation's largest banks and insurers that advanced artificial intelligence systems, specifically Anthropic's Claude Mythos model, pose a significant and accelerating cybersecurity threat. The Office of the Superintendent of Financial Institutions (OSFI) issued an urgent bulletin to senior executives on April 29, 2026, stating that such AI can dramatically shrink the time available to detect and patch software vulnerabilities, putting aging financial infrastructure at particular risk.

The warning was sent directly to chief technology officers, chief information security officers, and chief risk officers across the federally regulated financial sector. According to the regulator, AI models like Claude Mythos are designed to scan code for weaknesses at an unprecedented scale and speed. Banks that have reportedly had limited access to the tool discovered hundreds to thousands of previously unknown vulnerabilities in their systems, forcing emergency remediation efforts.

OSFI's alert is part of a coordinated global regulatory response that began building in early 2026. U.S. and UK regulators held meetings on the topic by mid-April, with...