Major US Banks Halt AI Transaction Access, Citing Security Risks and Liability Concerns

Charles Schwab CEO Rick Wurster announced on Friday that the firm is blocking unauthorized AI agents, including Meta’s Muse, from moving client funds. JPMorgan Chase CEO Jamie Dimon echoed this stance earlier in the week, warning against rogue agents accessing their banking infrastructure.

Schwab plans to allow outside AI tools to view balances through controlled Model Context Protocol servers, but only under strict rules. Wurster stated that clients should use Schwab’s own assistant, Charley, for interactions. This approach prevents external agents from executing transfers or payments.

Security context

The primary concern is clients handing passwords to agents that then act on their behalf. Wurster noted that liability for errors remains unclear in the US, where Congress is still debating AI agent responsibility. He receives weekly reports on these blocks, noting that major agents are stopped in the tens or twenties each week.

Industry alignment

JPMorgan Chase has similarly restricted third-party tools to viewing account data only, prohibiting them from sending money or paying bills. These moves coincide with US and European regulators questioning banks about their AI agent usage. Both firms aim to maintain control over financial transactions while limiting exposure to external automation.

Defender's checklist

MCP links have previously exhibited security flaws, with Google and JPMorgan fixing the same vulnerability. Organizations should monitor for updates regarding MCP server security. Banks are actively blocking unauthorized access attempts, indicating a broader industry shift toward restricting AI transactional capabilities to proprietary or tightly controlled systems.

Operational impact

Organizations must determine which internal processes rely on external AI agents for financial data access. Teams usually need to verify that third-party integrations comply with new read-only restrictions. Check whether your current AI tools can still retrieve necessary balance information without attempting transactions. Inform legal and compliance departments about the evolving liability landscape for AI actions. Ensure that staff are directed toward approved, vendor-specific assistants rather than generic agents. Review supplier contracts to confirm they align with these new banking security postures.

Background: AI agents

An AI agent is a software system that perceives its environment, makes decisions, and takes actions to achieve specific goals without constant human input. Unlike passive tools, agents execute tasks, browse data, and interact with other systems autonomously.

Read the full guide: AI Agents Defined: Autonomy, Risk, and Real-World Use

What to do and how to stay safe: Charles Schwab

• Verify that any AI tools connected to bank accounts are using approved, read-only interfaces.
• Avoid sharing banking credentials with unauthorized third-party AI agents or plugins.
• Monitor official bank communications for updates on supported AI assistants and security protocols.
• Review internal policies regarding the use of external automation for financial transactions.

Step-by-step guide: Implementing AI Usage Policies at Work: A Step-by-Step Framework

General security guidance from the InfoSec Ledger newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Can Meta's Muse move money from Schwab accounts?

No, Charles Schwab is blocking unauthorized agents like Meta’s Muse from moving client money.

Why are banks blocking AI agents from executing transactions?

Banks cite security risks and unclear liability for errors when clients hand passwords to external agents.

Will banks allow AI agents to view account balances?

Yes, Schwab will allow outside tools to check balances via controlled MCP servers, but not move funds.