UK and US Banks Tighten AI Agents’ Access to Customer Accounts
Major UK banks are restricting AI agents from trading independently in customer accounts, while some US financial institutions have blocked external agents from their platforms. Banks fear agents could act beyond customer authorization, prompting discussions about access rules and liability.
Starting the week of November 9, UK banks including HSBC, Barclays and Lloyds began tightening what AI agents can do on their platforms, prohibiting them from making transactions without the account holder’s direct involvement. The Telegraph reported the change.
Third-party AI tools can access accounts after users provide their login credentials, then shop, make payments or manage funds on their behalf. Banks worry that, once granted access, these tools could take actions beyond what users originally intended. When transfers are involved, banks also struggle to determine whether an action was properly authorized.
Illustration of AI’s impact on banking.（VietVance）
US financial institutions are also imposing restrictions. The Financial Times reported that Morgan Stanley has blocked external AI agents from accessing its banking platform, while JPMorgan Chase, Bank of America and Citi have taken similar steps. Bank of America said it is developing procedures to clarify responsibility for AI-authorized payments and believes transfers out of customer accounts still require human approval.
The UK’s Financial Conduct Authority is working with the financial sector and other regulators on rules for AI agents in finance. For now, banks decide whether to allow them to access accounts. Citing sources, the Financial Times said executives at some Wall Street financial institutions have discussed with Federal Reserve regulators how customers authorize automated tools and who is responsible when transactions cause losses. The Federal Reserve has not formally commented.
Source: VietVance
