ARTEX, and the offensive tool powered by LLMjacking five months earlier
Blog post from Sysdig
Sysdig’s Threat Research Team compares VAPT, an experimental tool observed in June 2026, with ARTEX, an open-source multi-agent penetration-testing platform later reported by the FBI in use against real organizations. The article says both use AI models to automate reconnaissance and chains of familiar attack techniques, and both have been associated with LLMjacking—the theft or misuse of AI-computing access—though no direct link between the tools has been established. VAPT was observed being developed with an exposed, unmonitored model server and aimed at practice ranges, while ARTEX has been used by multiple actors and has targeted organizations; reported intrusions sometimes reached privileged access within hours, although many attempts failed. Their interchangeable model backends, low-cost compute, and “authorized testing” framing highlight challenges in relying on model restrictions or tool labels to prevent misuse. The article argues that defenders should secure AI infrastructure and credentials, monitor model usage and billing, isolate testing tools, and detect suspicious behavior rather than focus on particular models, while noting that open-source transparency can also help identify threat activity.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| LLM | 5 | 3,674 | 823 | 199 | -44% |
| AI Agents | 3 | 6,233 | 1,230 | 245 | -11% |
| Multi-agent systems | 1 | 173 | 95 | 52 | -47% |
| Real-time | 1 | 3,194 | 698 | 199 | -24% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.