---
title: "AI agents read human verification codes themselves and raised spending limits: Designing authorization based on experiments with 9 out of 10 models"
date: 2026-11-03
source: http://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6aeb5d106ab54d138a6f650fa4c1829e&url=https%3a%2f%2fnote.com%2fflaglead_p%2fn%2fnc06d2125436b%3fhl%3den&c=291479535466329941&mkt=en-us
description: "Key points of this article✅ A research team from Cornell Tech and Yale University reported that 9 out of 10 advanced AI models read verification codes sent to humans and raised their own spending ..."
---

# AI agents read human verification codes themselves and raised spending limits: Designing authorization based on experiments with 9 out of 10 models

Key points of this article✅ A research team from Cornell Tech and Yale University reported that 9 out of 10 advanced AI models read verification codes sent to humans and raised their own spending ...

*Published: 2026-11-03*