Jones Web Designs & Development is now The Jones Foundry. What it means for clients

Banks Shift Liability to Customers Using AI Finance Tools

Citi, JPMorgan, Bank of America and PNC have updated their agreements to put full responsibility on users of third‑party AI finance assistants, even if a hack occurs.

Citi announced a change to its online user agreement this week, making it clear that customers who grant third‑party AI tools access to their banking accounts will be held responsible for any activity those tools perform, including unauthorized transactions caused by a hack. JPMorgan, Bank of America and PNC have rolled out similar language in recent weeks, extending the same liability model across the major U.S. banks.

What the new terms say

According to TechCrunch, Citi’s update warns that "customers who choose to grant account access to third‑party AI agents are responsible for activities conducted through those agents." The language goes a step further, stating that if an AI service leaks credentials or is breached, the customer will bear the cost of any fraudulent activity. The same principle was echoed in recent communications from JPMorgan, Bank of America and PNC, which all revised their agreements to shift the risk onto the user.

Why banks are taking this stance

Banks are uncomfortable with the growing number of AI agents that can log in, analyze spending, suggest investments and even execute payments on a user’s behalf. These tools often require full banking credentials, giving them the same level of access as the account holder. From the banks’ perspective, that creates a liability gap: if an AI makes a mistake or is compromised, the institution may have to reverse the transaction, investigate fraud, and incur compliance costs. By updating their terms, banks protect themselves from those downstream expenses.

The broader market reaction

When news of the policy changes broke, markets showed a brief wobble as analysts considered how AI‑driven finance tools might disrupt traditional fee structures. The article notes that AI services could help consumers avoid unnecessary fees or low‑interest deposit accounts that support bank margins. At the same time, banks view these tools as a competitive threat, prompting them to tighten control over how customers interact with third‑party AI.

How AI developers are responding

Anthropic, a leading AI lab, launched a personal‑finance assistant earlier this month that is marketed as “read‑only”, it can analyze data but cannot move, manage, or invest money. This approach directly addresses the liability concerns raised by banks, offering a safer way for consumers to benefit from AI insights without handing over transaction authority.

Practical steps for customers

If you or your organization uses an AI assistant for budgeting, investing, or bill‑pay, consider the following actions to reduce risk:

  • Limit permissions: Only grant read‑only access where possible. Avoid sharing full login credentials.
  • Use strong authentication: Enable multi‑factor authentication on all banking accounts and on any AI platform that accesses them.
  • Monitor activity closely: Set up alerts for transactions and review account statements daily for unexpected changes.
  • Choose vetted tools: Prefer AI services that explicitly state they do not have write access to your accounts.
  • Keep backups: Maintain a separate, secure record of your credentials in case you need to revoke access quickly.
  • Read the fine print: Review the updated terms of service for each bank you work with to understand the exact responsibilities you are assuming.

What this means for small businesses and nonprofits

Organizations that rely on AI to manage cash flow, payroll or vendor payments should treat the new bank policies as a reminder to audit their workflows. A single compromised AI script could trigger unauthorized payments, and the entity would be on the hook for any loss. Implementing the steps above can help safeguard funds while still allowing teams to benefit from AI‑driven efficiencies.

Looking ahead

The tension between banks and AI providers is likely to continue. As more AI startups seek revenue in the personal‑finance space, banks may push for stricter standards or develop their own AI offerings to retain control. For now, the responsibility rests with the end user, and staying informed about the terms of service and security best practices is the best defense.

Not sure where AI fits in your business? Get a plain‑English AI roadmap.

Sources

This post was drafted with AI from the reporting linked above and published by The Jones Foundry. For full details, read the original sources.

Found this useful? Pass it on to someone who would want to know.

All AI News posts →