Trusence Every claim has a source
Last updated 22 November 2026 Search Türkçe
Policy

Banks and retailers move to restrict consumer AI shopping agents

Liability for agent-placed purchases is still unsettled, and the institutions setting the rules are also building their own agents.

A brass clockwork automaton stands before a closed gate, with a stack of coins at its feet and one coin slipping off.
AI-generated illustration

Wells Fargo will roll out revised customer terms on Tuesday that warn people their use of any AI agent is at their own risk. Several other large banks have made similar moves this year: JPMorgan Chase banned third-party agents that log in with a customer's credentials, Bank of America cautioned customers against sharing login details with agents, and Citibank and PNC added agent warnings to their terms. Retailers are also pushing back, with Amazon blocking Meta's Muse from buying on its site and United saying personal bots will not be allowed to book flights for travelers. Researchers note that consumers are comfortable letting agents handle routine purchases like household supplies but not larger, more deliberate buys, and it is unclear who would bear the cost if an agent placed a mistaken order. American Express has indicated that some agent-caused errors may become eligible for reimbursement, while the banks are simultaneously building their own in-house agents.

Why it matters

For consumers, the practical effect is that a bank or retailer may now refuse to work with an agent acting on their behalf, and the terms they accept may put any resulting error on them. American Express has suggested some agent mistakes could be reimbursed, but that is not yet a general rule. Until responsibility is clearer, people who hand routine purchases to an agent are carrying risk that the companies providing the accounts and stores are shifting toward them.

Sources · 1