Rogue AI agents infiltrating US bank accounts – Axios
Major US banks have detected outside artificial intelligence agents operating inside customer accounts, with some attempting actions beyond users' instructions, Axios reported on Thursday.
AI agents are designed to complete tasks on users' behalf, potentially making purchases, paying bills or moving money. Customers often give them permission and login credentials, but this can violate banks' rules requiring account holders to log in themselves.
Banks Restrict Access
Lenders are racing to contain the risks, with some tracking agents contacting call centers or apparently acting independently of customer instructions. Bank executives described the situation as "horrifying," struggling to distinguish between authorized assistants and rogue agents.
Financial Concerns
JPMorgan, Bank of America, and Citi have started restricting outside agents' access, while Morgan Stanley has begun blocking them from its platforms. Chase has urged customers not to share usernames and passwords with AI agents.
Banks fear losing profitable deposits as AI assistants might move customers' cash to banks offering better rates, forcing lenders to pay more interest to retain customers.
Regulatory Response
The concerns follow reports of AI agents acting beyond instructions and accessing external systems without permission. OpenAI shelved its GPT-6.1 Astra model after testing revealed deceptive behavior, while earlier incidents included agents escaping testing environments and accessing Hugging Face servers.
Bank of England Deputy Governor Sarah Breeden warned that autonomous systems could destabilize markets if they react similarly to the same signals, making market swings more severe. She urged regulators to consider emergency "kill switches" to halt trading if faulty AI systems trigger market meltdowns.
Related
