Wells Fargo is quietly rewriting the rules of engagement for the agentic web. According to American Banker, the bank has updated its Online Access Agreement to include a new Section 15 that explicitly covers third-party AI agents. Effective November 24, 2026, this clause declares that any action taken by an agent under a customer's login is deemed initiated and authorized directly by the user. This is a massive shift in legal posture, moving from verification-based security to assumption-based liability.

The Burden of Proof Falls on You

The new terms place the burden of revocation squarely on the customer. If you delegate an AI assistant to manage bills or move funds and later decide to cut off access, you must retain and provide proof of that revocation to the bank upon request. Until Wells Fargo is formally notified, the customer carries full responsibility for the agent's actions. This includes unintentional errors and, critically, the consequences of manipulated inputs, a legal catch-all for prompt injection attacks where malicious strings trick an agent into harmful transactions.

A Strategic Divergence from Wall Street

This move appears to be the first documented liability transfer of its kind at a major US bank. It stands in stark contrast to the industry retrenchment reported by Axios and Reuters on October 22, where JPMorgan, Bank of America, Citi, and Morgan Stanley blocked third-party agents from accessing customer accounts entirely. While peers built walls to keep agents out, Wells Fargo built a waiver. The bank is choosing to allow agentic access but placing the risk on the only party that controls the agent: the account holder.

Key Takeaways

  • Effective Date: The new Section 15 goes live on November 24, 2026.
  • Liability Scope: Customers are liable for all agent actions, including prompt injection exploits, until the bank is notified of revocation.
  • Industry Context: This diverges from the blocking strategy adopted by JPMorgan, Bank of America, Citi, and Morgan Stanley.
  • User Action: Users must keep written proof of revocation for any AI agent authorized on their accounts.

The Bottom Line

Wells Fargo is betting that users will accept 'use at your own risk' for the convenience of agentic banking. If other banks copy this template, your AI agent's permissions become a critical security surface you own completely.