Censys finds 1,465 internet exposed ARTEX consoles

ARTEX, an open-source AI-driven offensive security framework released on GitHub in July 2026 by a developer going by Autumn-27, is spreading beyond the initial campaigns first identified. Written in Go with an embedded Next.js console, ARTEX provides an autonomous C2 platform that coordinates multiple AI agents and can be plugged into large language models.

CrowdStrike attributed ARTEX to a campaign against South Korean financial institutions in early October 2026, where the tool was used alongside Claude AI; Microsoft later linked ARTEX to a large-scale compromise of ConnectWise customer tenants, prompting IT management services provider to take several cloud services offline. In November, CERT-In warned that the threat group Crafty Falcon had targeted financial and cryptocurrency organisations with several AI-powered tools, including ARTEX.

Censys’ 5 November 2026 report by Silas Cutler expands the footprint. The researcher found 1,465 hosts exposing an ARTEX console to the public Internet, with around 650 in two blocks of identically configured hosting, leaving about 815 distinct deployments. ARTEX often shares infrastructure with other AI attack frameworks and traditional C2 tools such as Vshell, Cobalt Strike and Adaptix C2.

Censys also noted 32 ARTEX hosts with publicly accessible directories containing post-exploitation kits, stagers and implants, and 43 ARTEX hosts running self-hosted LLM gateways or reseller software. The study highlights an ecosystem around ARTEX, including forks and related tools like ScopeWeaver, ARTEX Pro, CyberStrikeAI, PentAGI, HexStrike AI and RedAmon, with operators deploying multiple agent frameworks in parallel, per Cutler.