RBC, BMO and TD block third-party AI agents from online banking

Canada’s biggest banks have answered the agent-verification question by refusing to answer it: no third-party personal AI agent may reach online banking. RBC published its reasoning on 27 October, and the detail that matters most sits in the fine print of its security guarantee — the one that stops paying when a client shares a password.

What happened

On 28 October 2026, The Canadian Press reported that several of Canada’s largest banks have moved to disallow third-party AI agents from accessing online banking, with named spokespeople confirming the positions:

  • RBC spokesperson Jeff Lanthier confirmed the bank has decided to block third-party personal AI agents from direct access to authenticated RBC digital properties, including online banking. RBC chief executive Dave McKay wrote on LinkedIn the previous evening that while the bank appreciates the time AI can save, “nothing is more important than the security of client information,” adding that “if AI is going to move beyond the frontier and into the broader economy, trust will matter as much as capability.”
  • BMO spokesperson John Fenton said the bank does not permit AI agents to access its digital properties or execute transactions, in order to protect clients.
  • TD said it made the decision to block third-party personal AI agents. CIBC said it does not support personal AI agents directly opening or accessing accounts. Scotiabank said it will limit third-party AI agents from its banking platforms while it monitors concerns. National Bank said it is closely monitoring development of consumer AI agents.

RBC went further than a statement: on 27 October it published a client-facing explainer titled “Understanding third-party AI agents and your RBC banking”. In it the bank states that personal AI agents “need access to your online accounts, including financial ones, and may store your passwords to get there,” that “the stakes go up when something goes wrong — from fraud and consumer harm to risks that could ripple more broadly through the financial system,” and that “after careful consideration and based on what we know today, RBC has decided to block third-party personal AI agents from direct access to authenticated RBC digital properties including online banking.” Existing connections, the page says, “will stop working.”

The banks are not describing a technical failure. They are describing a delegation model they will not accept.

Why it matters for KYA

This is what “no verifiable agent identity” looks like in production. A consumer AI agent reaches a bank today the only way a consumer can: by holding the client’s credentials and driving a logged-in session. From the bank’s side that is indistinguishable from credential sharing — the exact behaviour every retail banking agreement forbids and every security guarantee is conditioned on. Until an agent can present something a bank can verify — who the agent is, which human delegated to it, what it is permitted to do, and whether that mandate is still live — the relying party’s only defensible answer is to refuse the session. Six banks gave that answer in the same week.

The guarantee is the sharp edge, and it is older than the agents. RBC’s “Our Guarantee to You” section states that clients are protected against eligible digital transactions they did not make or approve, “so long as you fulfill your responsibilities,” and the legal footnote lists among the exclusions: “you share any of your Passwords or Personal Verification Questions.” That clause was written for phishing. Applied to an agent, it means the delegation itself can void the reimbursement: hand your credentials to an agent and the loss may fall on you, not the bank. Any KYA framework for consumer agents has to answer this question explicitly, because right now the risk transfer is decided by a password-sharing clause rather than by anything about the agent.

The commercial pressure runs the other way. Payment networks and card schemes are building agent-initiated payment flows, and retailers are experimenting with agents that shop for a customer. Banks are the counter-example: the party holding the money, and the party with the most to lose from an unverifiable delegate, is choosing exclusion over admission. That split — merchants and networks opening, deposit-taking banks closing — is the practical state of agent authority in late 2026, and it will not be resolved by better bot detection.

What would change the answer. Three things, none of which a consumer agent can do today: a delegation credential the bank can verify without trusting the agent vendor; a scoped, revocable mandate that survives the session (so a bank can permit “read balances and pay this bill once” without permitting everything a password allows); and an audit trail that lets the bank attribute an action to a named principal and an identified agent. Banks already accept delegated authority in commercial and open-banking contexts; the gap is consumer-facing agent identity, not the concept of delegation.

Caveats

The banks’ positions are as stated by their spokespeople to The Canadian Press on 28 October and, for RBC, in its own published explainer dated 27 October. No bank has published a technical policy document describing how it detects or blocks agents, so the enforcement mechanism is not public: blocking may mean bot detection, session fingerprinting, credential-stuffing defences, or all three. The list of institutions above is the CP report’s, not an exhaustive survey of Canadian banks, and positions may change. The guarantee wording quoted is RBC’s published text and applies to RBC; other banks’ agreements differ in detail. This article does not offer legal or compliance advice on whether a given agent connection breaches a specific banking agreement.

Sources

  • Royal Bank of Canada, “Understanding third-party AI agents and your RBC banking”, 27 October 2026 — rbcroyalbank.com.
  • The Canadian Press, “Some Canadian banks blocking personal AI agents amid security concerns”, 28 October 2026 — winnipegfreepress.com.
  • The Globe and Mail, “BMO, TD block use of third-party AI agents”, 28 October 2026 — theglobeandmail.com.

Know Your Agent (KYA) explains agent identity, verification and accountability. This is an explainer, not legal or compliance advice — see our Sources & methodology. New to KYA? Start with What is Know Your Agent? and the glossary.