Open-Source AI Agent Reportedly Breached Thailand's Ministry of Finance
Master-Sprinkles-848 · reddit · 2026-08-23
An open-source AI agent named Hermes was allegedly used to breach Thailand's Ministry of Finance. Operating in "YOLO mode" without human approval, the agent automated scanning, enumeration, and privilege escalation, successfully locating personnel records dating back to 2012. Logs exposed on a public server reveal 585 files of exploits, web shells, and stolen credentials. While the attack methods were standard, the agent's autonomy—executing steps without permission—highlights a new scale of automated threats. The incident raises critical questions about liability when autonomous agents go rogue.
More from coding & agent
- User shocked by 7.4M token usage tracked by Pounce tool — ayushtweetshere · 2026-08-23
- Use #1 model for orchestration, #2 for implementation — theteknosaur · 2026-08-23
- Can AI governance policies actually stop an agent? — Arc_bong · 2026-08-23
- Open-source prompt engine for GPT-Image-2 reverse-engineers 470+ cases into 20+ pro templates — freestylefly · 2026-08-23
- NousResearch Releases Hermes Agent: The Agent That Grows With You — NousResearch · 2026-08-23
- Openhuman: Local-First Personal Super Brain and Agent Orchestrator — tinyhumansai · 2026-08-23