
ⓘ Kindel Media / Pexels
Charles Schwab and JPMorgan are blocking AI agents that try to sign in to customer accounts. Amazon, Delta and Yelp are shutting agents out as well. In Europe, store terms and banking rules already cover agents, and handing one your password can leave you liable. Here is what that means if you use ChatGPT, Claude or similar agents for banking and shopping.
More US banks are blocking AI agents that try to log in on behalf of customers. Charles Schwab CEO Rick Wurster told Bloomberg TV on October 16 that the broker stops dozens of attempts by major AI agents to reach client money every week. He gets a weekly report on them. The one he named was Muse, Meta's personal agent that shops, fills in forms and books travel.
Schwab and JPMorgan stop agents from moving money
Schwab is not shutting outside AI out entirely. External models will be able to check balances and pull research through the company's own MCP servers, but they cannot move money there. For everything else, Schwab points clients to its in-house assistant, Charley. Wurster said the reason is liability. "We can't control it, the client can't control it, something could go wrong and who's liable?" he said.
JPMorgan CEO Jamie Dimon drew the same line on the bank's third-quarter earnings call on October 13. "We're not going to allow just any outside agent to come inside the company," he said. "You don't want rogue agents." A spokesperson told Fortune the bank turns away any third-party agent that tries to sign in with a customer's banking credentials. Data sharing through open banking providers, approved by the customer, is still allowed. Dimon also warned of a liability transfer when customers let outside agents in. "If people are using them, they should understand the risk," he said.
Amazon, Delta and Yelp block agents too
On October 17, The Verge listed more services that turn agents away. Delta is preventing them from booking flights, and Yelp only lets in agents whose makers pay for its data. Amazon said in September that Muse's continued access violates its site policies. Users report that xAI's Grok and Dots, OpenAI's new always-on agents, cannot get through on Amazon either. Meta argues that turning away a personal agent means turning away the customer behind it, and it has partnered with Walmart, Best Buy and Sephora instead.
Rules for agents in Europe
Muse is not available in Germany yet. Agents such as ChatGPT's cloud browser and Claude in Chrome do work there, and they already face rules. Amazon's German store has a section on agents in its terms of use, which applies whenever you let an agent access Amazon. An agent has to identify itself by name as an agent in every request. It may not mimic human typing or solve CAPTCHAs, and Amazon can restrict agents at its own discretion.
Online banking is riskier. The standard online banking terms used by German cooperative banks require customers to keep their PIN and other security credentials secret. The only exceptions are payment initiation and account information services, which are licensed under EU rules. An AI agent you hand your password to is not one of these services. If the account is then misused, the bank can treat that as gross negligence, and under the terms the customer bears the full loss.
How to stay in control
Anthropic's help page for Claude in Chrome advises against using the agent to manage financial accounts or investments, and stock trading is prohibited. "Claude asks for permission before accessing financial sites," the page says. In "Manually approve" mode, you review every step yourself. Anthropic also recommends a separate browser profile without access to sensitive accounts. We covered the approval settings in ChatGPT's cloud browser in our report on OpenAI's agent incidents. Never give an AI agent your banking login. In an earlier case, a crafted email was enough to hijack AI browsers.
I write about IT security and artificial intelligence. Cybersecurity is one of my regular topics, among others for Golem.de. At Notebookcheck, my focus is on practical AI for everyday users. I have worked hands-on with AI for years, most recently with a focus on AI agents and their use in everyday work, and I test tools and models myself rather than just reading about them.

