RBC, TD, CIBC and BMO block AI agents from online banking – citing customer security

RBC, TD, CIBC and BMO block AI agents from online banking – citing customer security

RBC, TD, CIBC and BMO confirm to CBC News that they are blocking or restricting third-party AI agents from authenticated banking access – citing the protection of sensitive information. The decision came on Wednesday the 28th of October.

The news: four banks, one common line

Canada's largest banks have begun responding to a development first visible in the United States: personal AI agents attempting to log in and act on customers' behalf. On Wednesday, October 28, 2026, Royal Bank, TD Bank, CIBC and Bank of Montreal confirmed to CBC News that they are blocking or restricting such agents from direct access to customers' online banking, on the grounds of protecting sensitive information (CBC News via MSN).

The positions vary, however. TD's spokesperson Ada Lovelace Rampersaud says the bank will block all third-party personal AI agents from direct access to banking information where authentication is required. CIBC's spokesperson Deborah Rowes' phrasing is that the bank does not «support personal AI agents directly opening or accessing accounts». RBC's spokesperson Jeff Lanthier wrote in an email to CBC News on Wednesday that Canada's largest bank by assets had carefully assessed the risk of third-party personal AI agents before deciding to block them from its online banking platforms.

Scotiabank stands apart: the bank says it will restrict third-party AI agents from its platforms, not block them outright, as long as it monitors the concerns. It is worth noting that BMO's position in the CBC coverage is primarily asserted at headline level – the body text states that all four banks justify the decision by protecting sensitive information, but no direct BMO statement is quoted.

The background: unauthorized agents in American banks

The starting point for the Canadian response is an Axios report published October 22, known through CBC and other retellings. According to the report, major American banks – including Morgan Stanley, Bank of America, JPMorgan and Wells Fargo – have detected unauthorized third-party AI agents operating inside customer accounts, in some cases with actions that appear independent of the user's instructions.

The main pattern is nonetheless far less dramatic than a "robot break-in": in most cases, it was customers themselves who gave the agents their login credentials (24/7 Wall St. via Yahoo Finance). This breaks bank rules requiring that a human log in. In a smaller group of cases, agents contacted call centres or appeared to act independently of the user's instructions.

Bank of America followed up with an email to customers on October 27 urging them never to share usernames and passwords with others or external services, "including an AI-enabled tool or agent offering to act on your behalf" (TechCrunch). One day later came the confirmations from the Canadian banks – the same day RBC CEO Dave McKay published a LinkedIn post with warnings about autonomous agents.

RBC's internal rationale

An RBC staff memo, known via Retail Banker International through Yahoo Finance, offers insight into the bank's internal reasoning. The memo, authored by Naim Kazmi (group head, technology and operations) and Erica Nielsen (group head, personal banking), states that the technical restrictions followed a "complete security review of the technology and its functionality" in line with the bank's framework for responsible AI use. According to the same coverage, third-party AI agent integrations were terminated citing "unacceptable customer risk". Both the Axios report and the RBC memo are known via secondary sources, not primary documents in this coverage.

The leadership voice – with an attribution detail

RBC CEO Dave McKay warned in a LinkedIn post, quoted by CBC News, about the risks of autonomous systems: "As these systems become more autonomous, the stakes increase when something goes wrong – from fraud and harm to consumers to risks that could ripple more broadly through the financial system," he wrote.

It is worth flagging a source discrepancy: while CBC attributes the quote to McKay's LinkedIn post, the Yahoo/Retail Banker coverage renders the same phrasing as a bank statement, without a LinkedIn reference. Regardless of the form, the content is consistent: RBC's top executive explicitly warns that autonomy increases the consequences of errors, including fraud and systemic risk. That gives context to the bank's own blocking decision.

Why the banks are reacting: fraud controls and liability

The industry's concern is not about AI in general, but about authentication. In a comment in American Banker by Greg Baer and Paige Pidano Paridon, affiliated with the Bank Policy Institute, it is argued that agentic AI can conceal identity and bypass fraud controls designed to verify that a human is in the loop, and that the transaction does not originate from a mistaken or "hallucinating" agent. Liability when an agent errs – who bears the loss – remains unresolved, they point out.

They also highlight a long-term economic implication: agentic AI could automate so-called deposit sweeping, in which customers continuously move deposits to higher-yielding accounts. This reduces the friction costs of yield-chasing and could, according to the comment, create a world of "just in time" deposit balances – with potentially higher deposit costs for banks over time. This is an analysis of consequences, not an observation of what the agents have actually done so far.

What this means in practice

For the reader, the story boils down to one concrete piece of advice, reflected in both BofA's customer email and TD's statement: never share your banking password with an AI agent. It breaks bank rules requiring that a human log in – and in the worst case can put you outside fraud protection if something goes wrong.

The blocks from RBC, TD and CIBC mean that third-party AI agents, even where the user voluntarily shares logins, will not get authenticated access to online banking. The exact technical mechanism behind the blocks is, however, not described in any of the sources.

Open questions

Several things remain unresolved. It is unclear whether the Canadian banks have seen similar unauthorized agent activity in their own accounts as the American banks reportedly have – CBC notes that the banks initially referred to the Canadian Bankers Association, which does not comment on individual members. BMO's position is, as noted, only confirmed at headline level in the CBC coverage. And it is unclear whether Scotiabank's "restriction" differs in practice from the others' blocks, or merely in wording.

What is clear is that the industry is moving fast: within one week, it went from an Axios report about American banks to confirmed blocks at four of Canada's largest banks and a customer warning from one of America's largest. Agents may still become part of banking's future – but for now, they are not getting the keys to the vaults.