AI Agent Vulnerability Assessment
Is this your project? Ask us to take it down
Gallery
About
As part of our red-teaming efforts, we recently configured GPT-6 Astra to test how easily an AI agent can bypass existing bot detection and human verification systems.
We built a tool that lets you run the same test on your own website: https://assess.poh.org
Enter your website URL, and we'll launch an AI agent to test your workflows, record its actions, and show how we evaluate the session.
Below are three recorded tests. In the recorded tests below, agents were able to complete CAPTCHAs, bypass existing bot detection mechanisms, create and verify accounts, access gated resources, and carry out end-to-end workflows intended for humans.
Hugging Face: https://assess.poh.org/test/2d75a3b713e2719e87a4b85c6b3aff70 (bypassed hCAPTCHA)
Cloudflare: https://assess.poh.org/test/0eb218ca6e2d7be993b689dd69a1cb20 (bypassed Turnstile)
YouGov: https://assess.poh.org/test/144f6d014fbf006b61ec6bc32ee8c3f4 (bypassed reCAPTCHA)
AI agents are ushering in a new age of online security, and we’re excited to hear your thoughts!
Comments (0)
No comments yet. Be the first to comment!
Related Products
Reswardo — Bóveda privada para tu legado digital
Reswardo — Production-ready encrypted digital legacy and continuity SaaS
SteelSuit
External web security scanner: TLS, headers, secrets & CVEs
Ad Blocker
FormCrab.com
Get messages while you hide your email
XCloak
MyVault — Personal Command Center
All-in-one personal vault for passwords, files, bills, and expenses.
ComingUp