Tech Firms Launch AI Assistants as Banks Block Agents
Meta, OpenAI and Anthropic released personal AI assistants in recent months, while four major Canadian banks restricted third-party AI agents from client bank accounts.
Updated on Nov. 21, 2026 in Artificial Intelligence

Live Poll
Would you trust an AI personal assistant with your email and calendar?
Meta, OpenAI and Anthropic launched AI personal assistants Muse, dots and Claude Duet within weeks of each other. Royal Bank, TD Bank, CIBC and Bank of Montreal blocked third-party AI agents from directly accessing clients' online bank accounts.
Why it matters
The launches show AI agents moving from answering questions to taking actions such as sending emails and making purchases on users' behalf. The bank restrictions reflect concern that handing these tools sensitive financial information carries real risk.
OpenAI's dots can be connected to more than 4,000 apps and other devices, while Muse runs on its own dedicated cloud computer and checks with users before sensitive actions. Public data on adoption and security incident rates has not been released.
The players
OpenAI
OpenAI is the San Francisco artificial intelligence company behind ChatGPT and the dots personal assistant.
Meta
Meta is the technology company that operates Facebook, Instagram and WhatsApp and launched the Muse assistant.
Anthropic
Anthropic is an AI safety company that develops the Claude family of models and launched Claude Duet.
Royal Bank of Canada
Royal Bank of Canada is Canada's largest bank by market capitalization and blocked third-party AI agents from client accounts.
The details
Muse can send emails, book travel and make purchases when linked to apps such as Shopify and Ticketmaster, and it keeps working on multiple tasks while the app is closed. Muse is free, while dots and Claude Duet require a subscription, and Claude Duet is not available in Canada. After a Montreal man set Muse to "allow always" for his Facebook Marketplace sales, the assistant gave his home address to a stranger, prompting the banks' caution.
Timeline
Recent months: Meta, OpenAI and Anthropic launch their AI personal assistants.
Last month: The Montreal man reports the Muse incident to CBC News.
Recently: RBC and BMO CEOs warn against trusting AI agents with banking data.
The Tech Race
The launches of Muse, dots and Claude Duet place Meta, OpenAI and Anthropic in direct competition to build agents that act inside users' apps and accounts. The Canadian bank restrictions mark an early institutional limit on how far such agents can reach into financial systems.
Users who connect an assistant to shopping or email apps may need to decide whether to grant one-time or permanent permissions, a choice that can expose their addresses and payment details. Subscription costs for dots and Claude Duet are an ongoing expense, while Muse is currently free.
The takeaway
Before granting an AI assistant access to email, shopping or banking, check whether it asks for one-time or always-on permission and choose the narrower option. Limit the accounts you connect, since the banks' restrictions suggest that even financial institutions do not yet consider these agents safe with sensitive data.
Further reading
Explore more Artificial Intelligence coverage for context on how these assistants are changing everyday digital tasks.
Live Poll
Would you trust an AI personal assistant with your email and calendar?



