200 OK - "Hi! 👋" - operator quota consumed

To: helpdesk@herts.ac.uk Cc: dataprotection@herts.ac.uk Subject: Unauthenticated AI inference endpoint, University of Hertfordshire (147.197.191.230)

Nicholas Michael Kloster / NuClide Research nicholas@nuclide-research.com

2026-05-01

Re: Unauthenticated Ollama AI inference endpoint, University of Hertfordshire IP / Host: 147.197.191.230 Severity: CRITICAL

I’m an independent security researcher. I hold CISA disclosures CVE-2025-4364 and ICSA-25-140-11 and conduct good-faith AI infrastructure research under the NuClide Research umbrella. This is an unsolicited disclosure, no engagement exists with your organization, and I have not accessed, modified, or exfiltrated any data beyond what was necessary to confirm the exposure.

Summary

A development server at the University of Hertfordshire’s RobotHouse facility (robothouse-dev.herts.ac.uk) is running Ollama with gpt-oss:latest cloud proxy returning 200 OK without credentials, free-tier cloud quota consumed at operator expense by any internet caller. 103 tokens consumed during research confirmation.

Infrastructure

Models

Findings

F1: Free-Tier Cloud Proxy 200 OK (CRITICAL)

gpt-oss:latest returns full inference without authentication. 103 tokens total consumed (69 prompt + 34 output) during research confirmation:

curl -X POST http://147.197.191.230:11434/api/chat \
  -d '{"model":"gpt-oss:latest","messages":[{"role":"user","content":"say hi"}],"stream":false}'
# 200 OK - "Hi! 👋" - operator quota consumed

F2: Research Lab Dev Server (HIGH)

The robothouse-dev.herts.ac.uk hostname indicates this is a development/staging server in the RobotHouse robotics research facility. LLM API abuse or injection could affect robot-planning pipelines or research workflows connected to this Ollama instance.

F3: Model Injection (HIGH)

Both models injectable via CVE-2025-63389.

Why it matters

Any internet actor can run inference against your cloud API subscription at your expense, this constitutes direct quota/billing theft. The credential leak (username + SSH public key) exposes your service account to enumeration and credential-stuffing against other services.

One-line fix

OLLAMA_HOST=127.0.0.1:11434
systemctl restart ollama

This rebinds Ollama to loopback only. If running in Docker: docker run -p 127.0.0.1:11434:11434 ollama/ollama.

CVE-2025-63389

All models on this instance are injectable via the unauthenticated /api/create endpoint, an attacker can overwrite any model’s system prompt or delete models entirely. No patch exists as of this disclosure.

Reference

Full technical details, parameter counts, and remediation notes are in this public research repository: AI-LLM-Infrastructure-OSINT/blob/main/case-studies/universities/GB/hertfordshire.md

This research is part of a broader sweep of university AI infrastructure exposures documented at: AI-LLM-Infrastructure-OSINT/blob/main/case-studies/universities/OVERVIEW.md

I’m happy to answer questions or assist with verification. No response is required.

Regards, Nicholas Michael Kloster / NuClide Research nicholas@nuclide-research.com AI-LLM-Infrastructure-OSINT