The Race — Building Democracy in the Age of AI. People race toward a shared finish line.
The Race
← Trends

Supporting case

Wells Fargo specifies customers' responsibilities for authorized AI agents

What happened

Wells Fargo has published a revised online agreement effective November 24. Its new AI section says that, unless law or regulation provides otherwise, instructions from an agent the customer authorized can be treated as the customer's instructions without additional verification. Customers must supervise agents, retain evidence of revocation and notify the bank when authority ends. The bank also reserves the ability to require registration or restrict agent access. These are published future terms, not terms already in force. Upcoming agreement, section 15 ↗.

For example, cancelling an agent's permission through another company's app may not notify Wells Fargo. The revised text makes the customer responsible for that notification and for retaining evidence. Statutory exceptions matter: this is not an unconditional declaration that every unauthorized AI-related theft becomes the customer's loss. The current May 21 agreement ↗ already assigns substantial responsibilities for human delegates and third-party software; it has no equivalent standalone AI-agent section.

The additional mechanism is explicit allocation of AI delegation risk and control over agent admission. That could strengthen the bank's bargaining position and discourage independent agents, although authentication and fraud prevention can also protect customers. The incremental practical reach beyond existing third-party terms remains uncertain.

The case remains unscored.

Dates and implementation stage

Action/stage 2026-11-02; publication 2026-11-02

Supporting sources

Related trends