Regulators convene to assess AI models capable of exploiting every major operating system

The Office of the Superintendent of Financial Institutions is assessing the potential impact of an AI model on the resilience of the financial system. The Bank of Canada and the Canadian Financial Sector Resiliency Group—including representatives from Canada's six largest banks, the parent company of the Toronto Stock Exchange, the federal finance department, and OSFI—met Friday to discuss cybersecurity risks raised by Anthropic PBC's Mythos model.

Mythos is capable of identifying and exploiting vulnerabilities in every major operating system and web browser. Because of this capability, Anthropic PBC has restricted the model from the general public and formed Project Glasswing to provide testing access to a group of major tech companies and JPMorgan Chase & Co.

Since 2023, OSFI has been tasked with evaluating whether banks and insurers have adequate policies to protect against security threats. The regulator stated it does not plan to issue short-term changes to its existing guidelines in response to this threat, but will continue sharing threat and mitigation information in coordination with the Canadian Centre for Cyber Security.