AIBound

Case study

A U.S. regional bank blocked an F-graded AI agent the same day

A U.S. regional bank had an AI policy and five approved tools, but no way to see what ran on employee devices. AIBound catalogued 143 AI agents, graded OpenClaw F on sight and blocked it the same day through the bank's existing EDR and SIEM, with no new tooling.

15 November 2026

Key takeaways

  • 143 AI agents catalogued, five sanctioned.
  • F grade given to OpenClaw on sight.
  • Products used: AISeeker, SkillGuard and AgentHarness.

The customer

A U.S. regional bank. Name and size are not cleared for publication.

The problem

How do you find an AI agent with standing access to payroll and banking systems?

The bank had an AI policy and five approved tools, but no way to know what was actually running. Employees had installed AI agents one at a time to get work done faster, and nothing in the environment showed which agents held access to sensitive systems or what they were told to do.

The solution

What they did

The bank connected AIBound agentlessly and acted on what it found the same day.

  1. 1 AIBound catalogued 143 AI agents on employee devices. Five were sanctioned.
  2. 2 AIBound graded OpenClaw F on sight, listed every ClawHub skill installed on it, and mapped the systems each agent could reach and the outbound channels it could use.
  3. 3 The block went out through the bank's existing EDR and SIEM. OpenClaw was blocked from launching, its messaging paths were cut and its stored credentials were flagged for rotation.

The results

What did the bank get from finding OpenClaw?

AIBound catalogued 143 AI agents at the bank, graded OpenClaw F on sight and blocked it the same day. No records left the bank before containment. The bank now sees which employees run which agents, and what each one can reach, with no new tooling and no manual investigation.

143

AI agents catalogued, five sanctioned

F

grade given to OpenClaw on sight

0

records exfiltrated before containment

Source: AIBound deployment results for this bank customer.

The quote

“We had an AI policy. What we didn't have was any way to know it was being ignored.”
CISO, U.S. regional bank

FAQ

Frequently asked questions

What is OpenClaw, and why did the bank block it?

OpenClaw is an autonomous AI agent with shell execution, file management, mailbox and calendar operations and links to eight messaging platforms. On some employee devices at the U.S. regional bank it held standing access to payroll information and a connection into banking systems, so AIBound graded it F and the bank blocked it.

How does AIBound grade an AI agent F?

AIBound looks at what the agent can actually do on the device, not at its reputation. For OpenClaw it listed every installed ClawHub skill, mapped the systems the agent could reach and the outbound channels it could use. That skill-level and permission-level detail turned an ordinary-looking productivity tool into an F.

Did the bank need new tools to block it?

No. AIBound pushed the block through the bank's existing EDR and SIEM, with no new tooling and no manual investigation. The agent was blocked from launching, its messaging paths were cut and its stored credentials were flagged for rotation. The bank already had an AI policy. What it lacked was a way to see it being ignored.

Stop the risky AI your employees already run.

AIBound blocks the high-risk AI on every employee device, in real time. The 14-day trial needs no proof of concept.

Book a Demo