BoC convenes banks on AI risk

The Bank of Canada brought major lenders and financial firms into a Friday meeting on April 10 to discuss cyber risks tied to Anthropic’s Claude Mythos model. (bloomberg.com) Bloomberg reported the meeting included the country’s biggest banks and financial firms, and came days after United States officials held a similar discussion with Wall Street leaders. (bloomberg.com) (cnbc.com) A spokesperson for Canada’s Finance Ministry said cybersecurity was discussed and that artificial intelligence and Anthropic’s new model were among the topics at the meeting, according to Reuters. (economictimes.indiatimes.com) Anthropic unveiled Claude Mythos Preview on April 7 as part of Project Glasswing, a restricted program for critical software defense rather than a general public release. Anthropic said the initiative launched with partners including Amazon Web Services, Apple, Cisco, Google, Microsoft, Nvidia and JPMorganChase. (anthropic.com) The basic concern is simple: a model built to find weak spots in software can also help attackers identify where to break in. Anthropic said Mythos Preview can find software vulnerabilities “better than all but the most skilled humans,” which is why access is being limited. (anthropic.com) (cnbc.com) Anthropic said it is committing up to $100 million in usage credits for the effort and described the model as capable of giving defenders a “head start” in securing critical infrastructure software. (anthropic.com) Canadian regulators were already building an artificial intelligence risk framework before this week’s meeting. In a September 24, 2024 report, the Office of the Superintendent of Financial Institutions and the Financial Consumer Agency of Canada said they were monitoring fast-moving artificial intelligence risks at federally regulated financial institutions. (osfi-bsif.gc.ca) The Bank of Canada has also framed financial stability work around new structural risks that can spread through the system, not just hit one firm at a time. In a 2022 explainer on climate risk, it said its job includes analyzing vulnerabilities that could affect the financial system more broadly. (bankofcanada.ca) For now, officials are treating Mythos less like a consumer chatbot and more like a powerful security tool with dual use. The Canadian meeting suggests central banks and finance ministries are moving to assess the risk before a real-world attack forces the issue. (bloomberg.com) (economictimes.indiatimes.com)