Big Banks Make Customers Liable for AI Agents
Citibank, JPMorgan Chase, Bank of America and Wells Fargo have revised their agreements to restrict third-party AI agent access to customer accounts.
Updated on Nov. 18, 2026 in Artificial Intelligence

Live Poll
Would you trust an AI assistant with your bank login?
Citibank updated its online user agreement with a new section on third-party AI agents, treating agent access and transactions as authorized by the customer. JPMorgan Chase, Bank of America and Wells Fargo have made similar changes to their agreements in late October and November 2026.
Why it matters
Defining AI agent activity as customer-authorized establishes in advance who absorbs losses from unauthorized or erroneous payments and account compromise. Customers who give AI tools their login credentials may bear responsibility for resulting account activity.
Four major US banks have revised their agreements in the past two months to restrict AI agent access, with Citibank ranking as the third-largest US commercial bank. JPMorgan Chase permits only read-only access through regulated open-banking platforms, while the stub does not say how liability would be resolved in a contested case.
The players
Citibank
Citibank is a major US commercial bank and the third-largest in the country by rank, which revised its online user agreement to address third-party AI agents.
JPMorgan Chase
JPMorgan Chase is the largest US bank by assets, which revised its Digital Services Agreement to bar third-party AI agents from accessing customer login information.
Bank of America
Bank of America is one of the largest US banks, which told customers that using AI tools with their credentials makes them responsible for account activity.
Wells Fargo
Wells Fargo is a major US bank headquartered in San Francisco, whose updated Online Access Agreement classifies AI agent-initiated transactions as customer-authorized.
The details
Citi's revised terms instruct customers not to share login credentials and classify any AI agent access or transaction as authorized by the customer. JPMorgan Chase's updated Digital Services Agreement bars third parties, including AI agents, from accessing login information, while allowing read-only analysis tools to operate through regulated platforms.
Timeline
Mid-October 2026: Major US banks began spelling out AI agent liability terms.
Late October 2026: JPMorgan Chase and Bank of America revised agreements restricting AI agents.
November 2026: Citibank changed its terms of service for AI agents.
November 24, 2026: Wells Fargo's updated Online Access Agreement takes effect.
Industry Context
JPMorgan Chase's allowance of read-only access through regulated open-banking platforms, while barring unregulated third-party credential sharing, marks the line the sector is now extending to autonomous AI agents. Citibank, JPMorgan Chase, Bank of America and Wells Fargo are drawing the same boundary in their own agreements.
Customers who connect AI tools to their bank accounts may now be responsible for any transactions those tools initiate. Reviewing whether a tool's access is read-only or transactional, and never sharing login credentials, is the practical step that limits exposure under these terms.
The takeaway
Before linking an AI assistant to any bank account, check whether it needs only read-only access or can move money, since the new terms treat agent-initiated transactions as your own authorization.
Further reading
For broader context on how these tools are changing consumer finance, see the Artificial Intelligence section.
Live Poll
Would you trust an AI assistant with your bank login?





