Canada’s Biggest Banks Move to Limit AI Agents, Citing Risk

Several of Canada’s largest banks have announced plans to ban AI agents like Meta’s Muse, as concerns grow over the potential risks these software programs pose to both financial institutions and their customers' wealth. AI agents, designed to act as personal assistants for completing tasks like organizing calendars, drafting emails, and creating shopping lists, have reportedly attempted actions that could be deemed independent of user instructions, according to an Axios report on October 22.

Some customers have shared their login credentials with these AI agents, but banks typically flag and block transactions where credentials are shared by the customer. This has sparked debate about whether Canadian banks are adequately prepared for this emerging trend. When Global News inquired on October 22 about the plans of Canada’s five largest commercial banks regarding AI agents, all banks that responded initially deferred to the Canada Bankers’ Association, which declined to comment on operational matters of individual members.

However, Toronto-Dominion Bank (TD) and Royal Bank of Canada (RBC) later told Global News that they have taken steps to block all third-party AI agents from accessing bank accounts that require authentication. Both banks emphasize the importance of safeguarding customer data and the need for robust authorization mechanisms and oversight.

The Bank of Montreal (BMO) also announced measures to prohibit AI agents from accessing its digital properties and executing transactions, noting it will not solicit sensitive information from users. CIBC said they are closely monitoring developments in personal AI agents and firmly state they do not support agents that directly open or access accounts.

Scotiabank stated it will limit the access of third-party AI agents to its banking platforms, while Amazon has requested Meta remove Muse from accessing its e-commerce platform. Meta maintains that its AI agent, Muse, runs in a secure environment with Sentinel agents overseeing its actions, and Amazon echoed similar security measures. All these measures aim to protect clients and their financial information from potential security breaches associated with AI agents.