After the United States, the Bank of Canada discussed the security risks of Anthropic's latest model with major financial institutions.

```

The Bank of Canada and the country’s major banks and financial institutions held a meeting on Friday to discuss cybersecurity risks triggered by Anthropic’s latest artificial intelligence model. A spokesperson for Canada’s Finance Minister François-Philippe Champagne confirmed the meeting took place on Friday.

This meeting closely followed similar actions taken earlier this week by US policymakers. Media reports on Thursday stated US Treasury Secretary Besant and Federal Reserve Chair Powell urgently convened Wall Street executives to discuss Anthropic’s Mythos and similar AI models.

The Canadian meeting was attended by members of the “Canadian Financial Sector Resilience Group,” which is composed of representatives from Canada’s six largest banks, such as Royal Bank of Canada and Toronto-Dominion Bank. Other members include the parent company of the Toronto Stock Exchange, the federal Ministry of Finance, and financial regulators like the Office of the Superintendent of Financial Institutions (OSFI).

The Canadian Financial Sector Resilience Group, led by the Bank of Canada, is a public-private partnership with the mission to “enhance the operational resilience of Canada’s critical financial systems.” A spokesperson for the Bank of Canada said: “We are aware of this issue and take cybersecurity very seriously.”

The Canadian Bankers Association, an industry organization representing dozens of institutions including the six major banks, did not comment specifically on the Anthropic model or the meeting. The association’s spokesperson said banks are managing AI-related risks through “long-standing, industry-specific regulatory requirements and internal frameworks.”

An OSFI spokesperson said in an emailed statement: “We are actively communicating with institutions to increase awareness of this issue and assess its potential impact on financial system resilience.” The regulator is staying in touch with banks regarding “the latest developments in advanced artificial intelligence models and their potential cybersecurity impacts.”

This meeting and OSFI’s response further indicate that global regulators are increasingly concerned that more powerful AI models could give rise to new types of cyberattacks targeting the financial industry.

Banks are increasingly forming dedicated teams to use AI to reduce costs and explore business opportunities, while also striving to manage the risks brought by this rapidly developing technology. Since 2023, OSFI has been tasked with assessing whether banks and insurance companies have adequate policies to guard against security threats and has issued guidelines on technology and cyber risk management.

The spokesperson said: “OSFI currently does not plan to adjust existing guidelines in the short term because of this emerging threat,” but will continue to coordinate with the Canadian Centre for Cyber Security, sharing threat information and mitigation measures, and continue to monitor related risks.

Anthropic said its Mythos model is a highly complex new system capable of identifying and exploiting vulnerabilities in mainstream operating systems and web browsers. Because of its excessive capabilities, the company decided not to release the model to the public, but instead set up a group called the “Glasswing Project,” providing test access to select major tech companies and JP Morgan Chase.

These companies will use the model to test their own products, find potential vulnerabilities, and provide feedback. Anthropic will then determine what safety guardrails are necessary for the technology. In the US, some Wall Street banks, including Goldman Sachs, Citigroup, and Bank of America, are conducting internal tests.

Risk Warning and DisclaimerMarkets carry risks; investment requires caution. This article does not constitute personal investment advice, nor does it take into account the individual investment objectives, financial situation, or needs of any particular user. Users should consider whether any opinions, views, or conclusions in this article are appropriate to their specific circumstances. Invest accordingly, at your own risk. ```