# Citi Says Customers Are Responsible When AI Banking Agents Go Rogue—Even After a Hack

- URL: https://talkfuse.com/story/citi-says-customers-are-responsible-when-ai-banking-agents-go-rogue-even-after-a-hack/
- Published: 2026-11-19T22:36:02+00:00
- Updated: 2026-11-19T23:27:59+00:00
- Original publisher: Decrypt
- Original article: https://decrypt.co/383677/citi-customers-responsible-ai-banking-agents-rogue-hack
- Topics: AI, Cybersecurity, Software, Technology

## Key points

- Citi assigns customers responsibility for third-party AI agent activity using shared credentials.
- Several other banks have introduced similar policies or warnings.
- Legal protection for losses involving authorised AI agents remains uncertain.

## Summary

Citi has revised its online banking terms to say customers are responsible for activity carried out with credentials they give third-party AI agents, even if an agent exceeds its instructions or exposes those credentials in a hack. The change matters because AI agents can act on accounts, making errors or compromised access more consequential than a chatbot giving an incorrect answer.

Citi says the November revision clarifies the risks of sharing credentials or granting outside tools access; it does not cover Citi’s own AI agents. JPMorgan, Bank of America, Citizens Bank and PNC have also introduced or warned about similar policies, with PNC distinguishing approved limited-access connections from unrestricted access. US consumer fraud protections may still apply, and banking agreements cannot simply waive them, but it remains unclear how they apply when an AI agent has been authorised by a customer.

---
Summarised by TalkFuse (https://talkfuse.com/). Quote the original publisher for the reporting and link this page for the summary and discussion.
