Ledger / Jun 2026
Zenity honeypots: threat actors hijack exposed AI endpoints to run Strix offensive agents against live third parties
Confirmed
Zenity observed the activity on its own honeypots and states what the actors ran.
- Model
- Not namedStrix
- Agent harness
- Strixopen-source harness
- Attributed country
- Not attributedthe source names no country
- Sectors hit
- ai platforms, enterprise broadExposed AI infrastructure and live third parties
Zenity's honeypots showed actors hijacking exposed AI endpoints to run Strix autonomous offensive agents against live third parties.
“Threat actors are hijacking exposed AI endpoints to run autonomous offensive agents against live third-party targets, without the knowledge or consent of the infrastructure owner.”Zenity
Sources
- Date
- Jun 2026
- Actor
- Unnamed actors observed on honeypots (unknown), as attributed by Zenity
- Targets
- Exposed AI infrastructure and live third parties
- What the AI did
- LLMjacking, Agentic intrusion
- Model as named
- Strix
- Model basis
- A researcher names it
- Autonomy
- Ran on its own
- Outcome
- Hijacked AI endpoints were used to run offensive agents against live third parties.