---
title: SOCRadar: 'KeyHunter' operator used Hermes Agent and OpenClaw bots to harvest 2,233 LLM API keys, map 247 open LLM gateways and run fake student sign-ups at 20 universities | ctxsec
url: https://ctxsec.com/items/socradar-keyhunter-leepsmarts-2233-llm-keys-ai-agents-university-fraud
description: "SOCRadar traced a Chinese-speaking actor, leepsmarts, through an open directory on their own server. Since July their pipeline scanned Vercel, Netlify, Railway, Render, Replit, Cloudflare Pages and GitHub for exposed .env and config files and collected 2,233 API keys from Anthropic, OpenAI, Google, DeepSeek and others (their tooling marked 12 valid). It also catalogued 247 exposed LLM gateways, 223 of them usable without a key. AI agents controlled over Telegram ran a model-release watcher, built the operator's dashboard, and automated fraudulent account creation at US universities. A cPanel CVE-2026-41940 exploit was staged against three universities, with no confirmed success. Harvard data came from an already-circulating 2025 dataset, not a new breach."
status: "confirmed research (SOCRadar Threat Research Unit, Nov 4; actor attribution high confidence for a single operator)"
published: 2026-11-04
---

> SOCRadar traced a Chinese-speaking actor, leepsmarts, through an open directory on their own server. Since July their pipeline scanned Vercel, Netlify, Railway, Render, Replit, Cloudflare Pages and GitHub for exposed .env and config files and collected 2,233 API keys from Anthropic, OpenAI, Google, DeepSeek and others (their tooling marked 12 valid). It also catalogued 247 exposed LLM gateways, 223 of them usable without a key. AI agents controlled over Telegram ran a model-release watcher, built the operator's dashboard, and automated fraudulent account creation at US universities. A cPanel CVE-2026-41940 exploit was staged against three universities, with no confirmed success. Harvard data came from an already-circulating 2025 dataset, not a new breach.

[← AI](/ai) · [markdown](/items/socradar-keyhunter-leepsmarts-2233-llm-keys-ai-agents-university-fraud.md)

confirmed research (SOCRadar Threat Research Unit, Nov 4; actor attribution high confidence for a single operator)

2026-11-04

### SOCRadar: 'KeyHunter' operator used Hermes Agent and OpenClaw bots to harvest 2,233 LLM API keys, map 247 open LLM gateways and run fake student sign-ups at 20 universities

- **Keep secrets out of anything a web server serves:** check that `/.env*`, `/wrangler.toml`, `/.vercel/output/config.json`, `/_next/env.js`, `/.streamlit/secrets.toml`, `/fly.toml` and similar paths return 404 on your deployments, and look at Wayback Machine snapshots too, since the actor checked old copies.
- **LLM keys:** set spend caps and alerts, scope keys per app, and rotate any key that has ever been in a repo or a client bundle.
- **Self-hosted gateways** (one-api, new-api, LiteLLM and similar): require auth on every route, including model lists and admin panels.
- **IOCs:** 209.112.91[.]74 (port 9997 dashboard), 173.231.39[.]244, 123.60.2[.]245, 14.47.118[.]194, 66.42.94[.]67; domains senhai[.]store, astrdark[.]cyou, xxtt[.]me, sajuai[.]xyz, vvbin[.]me. Full list in the report.

Sources: [SOCRadar: KeyHunter Operation: Harvard Alumni Data, Stanford Applications and LLM Key Theft (Nov 4)](https://socradar.io/blog/keyhunter-operation-harvard-alumni-stanford-llm-key-data/)