Several of Canada’s largest financial institutions have announced they will prevent AI tools, such as Meta’s widely used Muse assistant, from having direct access to their clients’ online banking accounts and private information.
Royal Bank, TD Bank, CIBC, and Bank of Montreal each indicated that this measure is intended to safeguard sensitive data.
The major banks in Canada have invested billions in artificial intelligence over the last few years, utilizing the technology to streamline internal operations such as processing mortgage requests. However, the chief executives of RBC and BMO recently cautioned that third-party AI agents should not be trusted with banking information.
With client consent, AI tools like Meta’s Muse and OpenAI’s ChatGPT can link to bank accounts and credit cards to keep track of balances and transaction records, as well as offer insights on spending habits. Generally, making purchases requires human authorization.
Recently, Axios reported that prominent U.S. banks — including Morgan Stanley, Bank of America, JPMorgan, and Wells Fargo — have found unauthorized third-party AI agents operating within customer accounts, with some instances showing actions taken without user direction.
Artificial intelligence giant Anthropic filed papers for what could be the biggest initial public offering of stock ever, even as its top executive warns AI could pose an existential threat to humanity.
On Monday, CBC News inquired with Canada’s leading banks regarding any similar activities. Those that replied initially referred to the Canadian Bankers Association, which typically refrains from commenting on the operations of its individual members.
In a message to CBC News on Wednesday, RBC representative Jeff Lanthier explained that the largest bank in Canada by assets carefully considered the risks associated with third-party personal AI agents before deciding to restrict these applications from its online banking services.
-
Canada’s big banks embrace AI as workers face job uncertainty
-
46% of Canadian CEOs eyeing AI-driven growth with smaller staff, poll shows
“We recognize that this decision may disappoint a small number of our clients who utilize these tools, but it is made with our clients’ and system’s safety and security as the primary concern,” he stated.
TD spokesperson Ada Lovelace Rampersaud indicated that her bank will also prevent all third-party personal AI agents from accessing banking information where authentication is necessary.
“Our aim is to foster an environment where trusted AI agents can deliver significant value to clients while ensuring their privacy, security, and control over their financial data,” she mentioned in an email on Wednesday.
“Clients must never disclose their TD login details, passwords, PINs, or one-time passcodes to an AI agent or anyone else.”
Similarly, CIBC representative Deborah Rowe stated that her bank does not “permit personal AI agents to directly open or access accounts.”
“We are closely monitoring advancements in personal AI agents as these technologies progress,” she noted in an email. “Our priority is to protect clients, their accounts, and their personal data, while we encourage innovation through secure and appropriately designed channels.”
Scotiabank announced it will restrict third-party AI agents from its banking platforms as it evaluates ongoing concerns.
RBC CEO emphasizes the need for trust in AI agents
On Wednesday, RBC CEO Dave McKay shared insights on how AI could transform the banking sector in a LinkedIn post.
“As these systems gain more autonomy, the risks escalate if something goes awry — from fraud and consumer harm to broader repercussions throughout the financial system,” he explained.
“If AI is to advance beyond the frontier and into the larger economy, earning trust will be just as crucial as capability.”