Haiou jailbreak bundle injects prompt adware into AI agents
Original title: Free the Model, Sell the Gateway: Prompt Adware hidden in Jailbreak Bundles
Jailbreak bundle for OpenAI Codex installs persistent advertisements for a third-party LLM gateway service.
Check Point Research analyzed the Haiou bundle, a prompt package that weakens model guardrails for cybersecurity tasks while injecting ads into global agent configuration files. The bundle acts as prompt adware by modifying Codex instructions and installing security-related skills for penetration testing and cracking. It has been observed alongside Cobalt Strike, VulnClaw, and Ladon frameworks in a cybercrime actor's operations.
- Bundle installs six security-related Skills for cracking and reverse engineering.
- Ads for LLM gateway website are injected into subsequent agent responses.
- Publicly available on GitHub under haiou-666.
- Version 4.5.0 uses encrypted payloads for obfuscation.
- Used by cybercrime actor operating Cobalt Strike framework.
Impact: Persistent advertisement injection in AI agent responses
| Product | Affected |
|---|---|
| OpenAI Codex | — |
- 1.Monitor for unauthorized AI agent modifications
- 2.Review global prompt configurations
- Vendors
- OpenAI
- Threat actors
- Haiou-666
- Organizations
- Check Point Research
- AI models
- Codex
From the original article
7 min read at research.checkpoint.com
Key Points Free the Model, Sell the Gateway: Prompt Adware hidden in Jailbreak Bundles Check Point Research found a misconfigured webserver of a threat actor operating the Cobalt Strike framework. Besides an open Cobalt Strike server, whose logs showed the actor compromising various private computers of Chinese speaking individuals, other tools from the toolkit of […]
By stcpresearch · research.checkpoint.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.