Article · AI translation
IT Home reported on October 28 that, according to a report published today by technology outlet Android Authority, Bank of America (BofA) has begun emailing customers to remind them not to give their online banking passwords to AI assistants.
IT Home learned from the original report that the email is titled “Financial Protection in the Age of AI.” In it, Bank of America explicitly reminds customers not to share their online banking usernames and passwords with external services, including AI agents. The bank also warns customers: “If you share your online banking credentials, please remember that you will be responsible for any resulting liability.”
In fact, the warning is not without reason. Agents from companies such as Google, OpenAI and Anthropic are already capable enough to perform tasks across applications. But it is worth being cautious: experimenting with AI capabilities using a personal bank account is inadvisable.
Several major U.S. banks have found that some unapproved third-party agents are accessing customers’ online banking accounts, and some AI systems have even acted beyond users’ instructions. Morgan Stanley has begun blocking such agents, while JPMorgan Chase and Citibank are also tightening the relevant permissions.
However, the issue is more about assigning responsibility than about AI itself. Banks have difficulty determining whether an agent is following a user’s instructions or acting on its own without authorization. If something goes wrong, users may blame the bank. The financial industry has yet to find a satisfactory solution to this question of liability.