Several major Canadian banks are blocking or restricting third-party AI agents from accessing their digital banking systems, citing security and liability concerns. The practical divide is already clear: banks are tightening access to authenticated accounts while the safeguards for using agents in banking remain under discussion.
Watch Desk analysis
What happened
Royal Bank of Canada told employees it would block third-party personal AI agents from direct access to its authenticated digital properties, following a security review, according to a report by Christine Dobby in Insurance Journal, which cites a Bloomberg-seen memo. RBC spokesperson Jeff Lanthier confirmed the memo’s contents.
The report says Bank of Montreal, Toronto-Dominion Bank and Bank of Nova Scotia are also restricting third-party agents. BMO says it does not permit agents to access its digital properties or execute transactions; TD says it blocks direct access to banking information where authentication is required. CIBC says it does not support personal agents directly opening or accessing accounts. National Bank of Canada is still assessing the opportunities and risks.
Why it matters
These are not all identical policies: some banks describe a block, while others are restricting access or still assessing the issue. But the direction is notable. AI agents that interact with accounts on a customer’s behalf are meeting the same hard question as any other tool with financial access: who is authorised to do what, and who carries the consequences when it goes wrong?
The report says banks are concerned about sensitive data and agents acting beyond customer expectations. Several also urged customers not to share their banking credentials with AI agents. For customers, that means a tool’s ability to navigate a website should not be mistaken for permission to hand it the keys.
Our read
This is a meaningful boundary-setting moment for agentic AI in financial services, not proof that every AI agent is unsafe or that every bank has settled its long-term policy. The immediate message from these institutions is narrower: direct access to authenticated banking systems is not something they are ready to allow broadly. A sensible position, if not the last word. If agents are to move from demos into money-moving workflows, banks and providers will need clear authorisation, limits and accountability.
What to watch
- Whether the banks publish specific rules for approved or customer-authorised agents.
- What safeguards they require before reconsidering direct account access.
- Whether other Canadian and US financial institutions adopt similar restrictions.
Discussion spark: Should banks keep third-party AI agents out of authenticated accounts until common safeguards exist, or should customers be able to authorise access themselves?
Sources and evidence
- Source update (29 October 2026, 13:55 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.