SecOraLens
HighPhishing

ARTEX AI Pentest Tool: What Security Teams Need to Know

SOCRadar·Source published ·1 min read

What happened?

ARTEX AI has moved from a GitHub research project to the engine behind some of the fastest-moving agentic AI cyberattacks of 2026. ARTEX is an autonomous penetration testing framework that uses coordinated Large Language Model (LLM) agents to run reconnaissance, exploitation, privilege escalation and data theft with little human input.

Key details

  • ARTEX AI is an autonomous penetration testing framework using coordinated LLM agents for full kill chain operations.
  • Vapor Panda ran a single ARTEX controller against over 200 organizations in 14 countries over 11 days starting October 17, 2026.
  • ARTEX agents achieved privileged cloud access in under three hours in multiple Storm-0940 intrusions.
  • A South Korean criminal ring used ARTEX to breach about 60 organizations and steal 8.2 million records for extortion.
  • ARTEX leverages multiple LLM backends including DeepSeek v4.1-flash, GLM-5.3, Grok 4.6, and an unidentified commercial model.
  • Google and Mandiant observed ARTEX agents pulling over 110,000 Active Directory objects and using Kerberoasting and DCSync for domain compromise.

Affected

  • South Korean banks including Shinhan Bank, KB Kookmin Bank, and Hana Bank.
  • WorkOS internal sales and marketing API.
  • Government ministries in Latin America, Europe, Asia, and Middle East.
  • US defense contractors, semiconductor, pharmaceutical, financial, nuclear sector organizations, and think tanks.

Fixes and mitigations

  1. ConnectWise disabled seven compromised Cloud Backup application identities and began phased restoration via new application.

What to look for

  • Burst of 27 automated events in 23 seconds indicating AI-driven attack tempo.
  • Password spraying at one-second intervals against software engineer accounts at IT vendors.
  • Probing of Okta, Entra ID, and Cognito identity providers for supply chain access.
  • Use of unsigned SAML assertions chained with XXE flaws to query cloud metadata services.

Summarised by SecOraLens from SOCRadar. Always read the original source before acting. How SecOraLens verifies advisories.

Back to all advisories