Canada’s federal banking regulator has warned the country’s largest financial institutions about the risks posed by Anthropic’s Claude Mythos and other advanced AI systems, according to a report by Reuters. In an April email obtained by Reuters, the Office of the Superintendent of Financial Institutions (OSFI) cautioned that frontier AI models could increase cyber threats and compress the time banks have to identify and fix the vulnerabilities. The email, sent to chief technology, information security, and risk officers across Canada’s financial industry, outlined practices to enhance risk identification and response. “Advanced artificial intelligence models, such as Anthropic Claude Mythos, significantly compress the timeframe for effective risk mitigation,” OSFI wrote.Following Reuters’ inquiries, OSFI published a public bulletin emphasising its technology-neutral, risk-focused approach. The regulator said its concern is not the technology itself but how institutions govern and manage risks associated with its use.
Global concerns over Mythos
Cybersecurity experts have described Mythos as extremely capable at finding and exploiting vulnerabilities, raising alarms for legacy banking systems. The warning comes after U.S. regulators, including Treasury Secretary Scott Bessent and then-Federal Reserve Chair Jerome Powell, convened urgent meetings with bank CEOs earlier this year to discuss similar risks.
Banks respond with AI defenses
Canada’s big six banks — Royal Bank of Canada, TD Bank, BMO, Scotiabank, CIBC, and National Bank — have all disclosed AI initiatives ranging from chatbots to internal tools. RBC’s AI Group Head Bruce Ross said Mythos underscores a shift in the cyberattack landscape: “The way we’re dealing with it is building our own AI defenses… we’ll continue to do that.”
Government access and industry impact
The Canadian government has said it has access to Anthropic’s Project Glasswing, which enables companies to use Mythos, though it remains unclear which banks are deploying it. The Canadian Bankers Association noted that institutions have invested heavily in cybersecurity and comply with OSFI’s robust requirements for risk management and incident reporting.
Anthropic just made an admission on Claude that may scare many of companies
Recently, Anthropic revealed that its AI model Claude uses a small enterable workspace to hold and manipulate ideas without expressing them in words. The said that this structure, dubbed ‘J-Space’, shows intriguing similarities to how humans consciously access thoughts. According to a report by Axios, in a video demonstration, Anthropic explained, “We can see Claude silently perform reasoning steps in its head—noticing bugs in code, identifying images, and more.” The J-Space operates separately from the “chain of thought” reasoning Claude shares with users, allowing the model to plan strategies unrelated to its immediate task.Anthropic’s findings also highlight a division between deliberate reasoning and the larger volume of automatic computation beneath it. In the research paper the company used a word “conscious” more than 200 times, though it stopped short of claiming Claude is conscious. The discovery adds fuel to ongoing debates over machine consciousness and whether advanced AI systems are approaching AGI.