Several banks are warning customers that they could be held responsible for problems caused by “AI banking agents,” including cases where those agents act incorrectly after account access is compromised. The warnings focus on customer liability and how banks treat use of outside or connected AI tools in banking workflows.

Citibank, in particular, is reported to have updated its customer terms to place responsibility on customers for actions taken by external AI agents, including scenarios involving leaked logins. Business-focused coverage also describes a broader pattern across multiple banks: institutions are highlighting risks in letting AI systems handle financial tasks and cautioning that mistakes may not be covered like ordinary bank errors.

Across outlets, the emphasis differs between a specific example (Citibank’s updated terms) and a wider survey of similar bank warnings. Both frames point to the same central issue: what happens when AI-driven banking activity goes wrong, and how liability is allocated when access credentials or agent behavior are involved.