Citi tightens AI agent policy for retail banking customers
Major U.S. banks are increasing restrictions on the use of third-party AI agents in personal finance as automated tools gain access to consumer accounts and payment functions. Citi now says customers bear the risk not only for transactions initiated through those tools, but also for losses tied to credential leaks or hacks involving the agents.
Highlights
- Citi updated its online user agreement this week, shifting liability for third-party AI tool activity on accounts entirely onto customers.
- The new policy holds customers responsible for unauthorized access or leaked credentials if an AI agent is hacked or exposes their information.
- JPMorgan, Bank of America, PNC, and Citizens implement similar changes as banks push back against AI agents, potentially slowing sector-wide adoption of AI in personal finance.
Updated account terms shift liability to users
As reported by TechCrunch, Citibank has updated its online user agreement this week to state that customers who grant third-party AI tools access to their accounts are responsible for activity conducted through those agents. The change applies even when an automated tool makes an intentional or unintentional transaction, reflecting growing concern among banks about unreliable outside systems interacting with core banking platforms.Citi goes further than some peers by warning that customers are also responsible if an AI tool leaks their credentials or is hacked, allowing attackers to access the account. A Citi spokesperson says the bank wants customers to understand the risks of sharing banking credentials or authorizing third-party AI tools to use their accounts, and says the revised language is intended to clarify those risks.
Broader banking pushback hits AI finance tools
JPMorgan and Bank of America, the two largest banks in the country, announced similar policies last month, while PNC also changes its terms today. Citizens likewise emails customers this week to warn that they are on the hook for actions taken by AI assistants, according to an email seen by TechCrunch.The policy moves come as AI companies seek new revenue opportunities in personal finance, where tools promise to analyze spending, advise on investments and handle bill payments or other transactions. But banks appear wary of accepting liability for errors or security failures created by third-party agents, a stance that could slow adoption of AI-driven financial services across the sector.
The reaction also fits a wider pattern across service industries, where AI developers promote agents as a new interface for consumers while incumbent providers resist giving those systems broad operational freedom. How that tension develops may shape whether AI investments in consumer finance translate into durable business gains.
Earlier, we discussed JPMorgan and Hyundai Card’s AI agent testing for simulated FX transactions in corporate treasury.