Banks and AI Agents: A Delicate Balance
Banks are banning AI agents over security... and competition concerns
AI agents are set to automate everyday banking tasks, but banks aren’t ready to cede control of customer relationships. Four banks in the UK have moved to block or restrict third-party AI agents from accessing customer accounts, especially where transactions can be made without customer involvement.
There’s currently no evidence of AI agents accessing UK bank accounts, but lenders are wary of potential issues if automated systems mishandle customers’ money and the resulting liability. The banks also have unspoken concerns, as Edgars Nemše, chief executive of GenLayer Foundation, explains. Routine banking tasks like bill payments, money transfers, and finding better rates could all be handled by agents. This would enable agents to seamlessly offer customers better deals elsewhere as soon as they’re available online, making the banking landscape even more competitive.
“For plain accounts, that removes the switching costs and makes the space even more competitive,” says Nemše.
The Risk of Agentic Bank Runs
In September, Apollo chief economist Torsten Slok warned about a potential “agentic bank run”. If AI agents collectively move household cash from low-yield checking accounts to higher-yielding alternatives en masse, it could drain banks of the low-cost funding they rely on for lending. The speed and scale of such an event “would be a problem for the entire financial system”.
JPMorgan Chase CEO Jamie Dimon has also expressed concerns, stating on an October earnings call, “you don’t want rogue agents”.
The Future of AI in Banking
Despite short-term efforts to hold back AI, it’s expected to revolutionize banking within a few years. Neeraj Makin, group head of strategy analytics at Emirates NBD, predicts agents will enable 24/7 banking and near-instant payments by 2036.
“Forget about waiting for a payment for 24 hours or inter-country payments taking three days… all of that will be in seconds, atomic,” he says, with AI agents handling most financial decisions.
Shifting Customer Interactions
Ferdinand Dabitz, CEO of Augustus, an AI-native banking company, foresees a similar shift in how customers interact with financial institutions.
“Every consumer will have 10s and every business will have 100s of agents talking to their bank,” says Dabitz. A human interface will still be crucial for managing these agents and providing trust, especially in commercial banking.
Building Agentic Frameworks
Emirates NBD’s developer experience team argues that financial institutions should create their own agentic AI frameworks, rather than allowing individual teams to develop agents without common controls. Its Leap framework governs agent authentication, data access, and actions, with audit trails, monitoring, and human oversight built in.
However, there are significant concerns about the reliability of agents to make critical financial decisions independently. A recent Bank of England and FCA survey revealed that while 80% of responding firms had deployed AI agents, only 2% used agents with broad autonomy, citing data security, third-party dependencies, and risks as key concerns.