Royal Bank of Canada's chief is considering a ban on Meta's autonomous agent, joining a growing list of institutions wary of AI touching client money
The chief executive of RBC is considering blocking Meta ’s Muse, the personal AI agent that has spent the past several weeks testing how far big institutions will let software act on a customer’s behalf.
Nothing is final yet. But if RBC follows through, it would join a fast-forming coalition of financial giants who have looked at Muse and decided the answer to “can I touch the money?” is a firm no.
Meta launched Muse on September 8, 2026. It is pitched as a personal AI agent that can take autonomous actions across apps instead of only answering questions.
Muse comes with a free tier. Paid subscriptions run $20/month for the Power plan and $100/month for the Maximum plan.
For financial features, Muse integrates with Plaid to pull in account data and generate insights. Banks have allowed that connection only in a limited, read-only form, with transactional actions strictly off the table.
In practice, Muse can look at your balance. It is not supposed to move it.
The consumer response has been strong. Muse climbed to the top tier of app store rankings shortly after launch and reportedly drew millions of downloads, which is exactly why regulated industries are paying close attention.
The pushback started well before RBC entered the conversation.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Amazon blocked Muse from its platform in September 2026, citing policy violations tied to unauthorized access, credential handling and data scraping.
Then the banks weighed in. During JPMorgan Chase ’s Q3 earnings call on October 13, 2026, CEO Jamie Dimon said the bank prohibits outside AI agents because the risks are too high, framing the policy as protection for both clients and the institution itself.
Three days later, on October 16, 2026, Charles Schwab CEO Rick Wurster confirmed that his firm blocks attempts by AI agents like Muse to move client money every week. According to Wurster, Schwab sees an average of tens of attempts per major agent.
When a customer moves money themselves, responsibility is clear. When an AI agent moves money on a customer’s behalf, and something goes wrong, the question of who pays becomes messy very quickly.
Banks are also discussing the broader risks of AI agents with regulators. The focus of those conversations remains on shielding clients from fraud and errors that could stem from automated interactions.
Credential handling adds another layer. Amazon’s decision specifically flagged how Muse managed login information, and financial institutions treat credential security as close to sacred.
For Meta, the friction exposes a gap between consumer enthusiasm and institutional permission. If banks and retailers keep the gates closed, a $100/month Maximum subscription is a harder sell if the agent cannot complete the tasks users care about.
Plaid sits in an interesting spot here. Its integration is what gives Muse its financial insights, and the read-only arrangement shows how data-sharing infrastructure can serve as a middle ground between full access and an outright ban.
What to watch next is whether RBC formalizes a block and whether other Canadian banks follow its lead. Also worth tracking: the outcome of the ongoing discussions between banks and regulators. Clear rules on agent liability could either unlock transactional access or cement the current lockout for years.