Canadian banks block third-party AI agents over security and liability concerns

Royal Bank of Canada, Bank of Montreal, Toronto-Dominion Bank and Bank of Nova Scotia have restricted third-party artificial intelligence agents, including Meta’s Muse, from accessing their digital banking systems. The banks say this step follows security reviews and aims to reduce unacceptable customer risk and potential liability exposure. RBC’s group heads of technology and personal banking told staff the decision was based on the safety and security of clients and systems. Other banks, including Canadian Imperial Bank of Commerce and National Bank of Canada, are either blocking such access or monitoring the issue closely.

Bank spokespeople note that only a small proportion of clients use these AI agents, but the risk of unauthorised access and data loss is significant. Several banks urged clients not to share credentials with AI agents and said they are interested in AI innovation but need stronger safeguards before allowing access. US banks have taken similar steps, with Charles Schwab’s CEO recently confirming active blocks on unauthorised AI agents.

This move reflects growing concern in financial services about the risks of third-party AI tools acting beyond customer instructions or exposing sensitive data. Senior executives should assess where their own AI policies may need to be tightened, especially as client demand for automation grows.