Citi, PNC, Citizens tell customers they're liable for third-party AI agent actions, even if hacked
jomic01 · reddit · 2026-11-19
Citi updated its online user agreement this week: anything a third-party AI agent does in your account counts as authorized by you — even if the agent ignores your instructions or its credentials are leaked and hacked. Customers are also expected to monitor agent activity, and Citi can cut off an agent's access without notice. The bank's own AI tools are exempt.
PNC and Citizens made similar changes this week, and per TechCrunch, JPMorgan and Bank of America did the same last month — a coordinated shift of AI-agent risk onto customers.
More from coding & agent
- Camoufox, the anti-detect browser for agents, vanishes from GitHub — citrusalex · 2026-11-19
- CI usage explodes 12x in a month: 51K to 623K runner minutes as agents drive workloads — aniketmaurya · 2026-11-19
- Matt Pocock: Much of my AI Coding Crash Course ports directly to non-coding work — mattpocockuk · 2026-11-19
- bclavie: Neon is one Claude Code-level in-house harness away from a step-change in feel — bclavie · 2026-11-19
- Antigravity lands in T3 Code nightly as Google says it won't ban Gemini subscribers — vista8 · 2026-11-19
- GateRook: one desktop app replaces 100+ MCP server setups with a local gateway, scoped permissions and approvals — JeremyCMorgan · 2026-11-19