Schwab says it stops AI agents 10 to 20 times each week

Banks tighten their rules against AI agents that use customer accounts. Charles Schwab chief executive Rick Wurster told Bloomberg that large agents are stopped 10 to 20 times each week. Ally Financial says that a customer who shares a login with an AI service is responsible for what that AI does. Wells Fargo says it took steps to stop agents that are not approved.

Why banks stop agents

Ted Miracco of the mobile security company Approov says banks cannot tell a legitimate agent from an attacker with the same credentials. He says that to stop agents is the most reasonable defense until agents have verifiable identity, scoped permissions and audit trails.

What users can do

Experts recommend four steps:

• Examine the permissions that you gave to agents, and remove more than the task needs.
• Change passwords that you shared with an AI service.
• Set multi-factor authentication and transaction messages.
• Look at account work each day.

What is not known

The count of stops comes from one executive. It can increase as bot use grows. The source gives no data on the cost of agents to users.

Insight: Banks block first, so agent builders need proof of identity

Five UK banks banned agents before any attack, and Citigroup and Bank of America limit them. Schwab offers its own tool, Charley. This looks like banks that want to keep the customer. They stop external agents and offer their own. An agent with proven identity can get access first. Teams that make agents must plan for refusal.