Major US banks have reportedly identified third-party artificial intelligence (AI) agents accessing customer accounts without authorisation. Some of these agents have allegedly attempted to carry out actions that seem to go beyond the instructions provided by users, sources told Axios.
The development has raised serious concerns among banking leaders, who are working to tighten restrictions on access to their systems. Industry and banking executives told Axios that the rapid pace of AI advancement is making it difficult for financial institutions to keep up.
Several banks are also drawing up contingency plans to address worst-case outcomes, including “day after” scenarios. These measures involve preparing to quickly secure customer accounts or restore them if an AI-related incident causes disruption, sources said.
Another major challenge for banks is identifying whether an account is being accessed by a human or an AI agent. Executives said it remains difficult to establish whether such agents are operating with customers’ consent or acting independently without their knowledge, raising fears of rogue AI activity.
