Dev.to Security ๐Ÿ” Cybersecurity ๐Ÿ‘ 0 ๐Ÿ“– 5 min read

FBI FLASH: ARTEX AI Pentest Framework Takes Intrusions From Targeting to Privileged Access in Under Three Hours

TL;DR what: The FBI issued a FLASH advisory warning that multiple threat actors have used ARTEX, an AI-driven autonomous penetration testing framework, to automate cyberattacks worldwide since at least August 2026.

TL;DR

  • what: The FBI issued a FLASH advisory warning that multiple threat actors have used ARTEX, an AI-driven autonomous penetration testing framework, to automate cyberattacks worldwide since at least August 2026.
  • impact: ARTEX was used in the ConnectWise Cloud Backup supply-chain attack, which hit at least 162 Microsoft 365 tenants and targeted more than 2,250, and in attacks on South Korean financial institutions that exposed data on about 67,000 people.
  • fix: There is no patch because ARTEX is an attack tool; the FBI's 13-page advisory includes IOCs and hardening guidance, and MalwareMustDie has released an open-source detection signature pack and a confidence-graded detection playbook.
  • who: Organizations in financial, government, technology, defense and healthcare sectors are at risk, especially those with exposed partner portals, legacy services, third-party integrations and cloud application permissions.

The FBI says an open-source AI hacking framework called ARTEX is letting threat actors move from initial targeting to privileged access in under three hours. In a FLASH advisory released Friday, November 14, the bureau said multiple groups have weaponized the tool since at least August 2026 against organizations in the financial, government, technology, defense and healthcare sectors. One of those campaigns was the October ConnectWise Cloud Backup supply-chain attack. In it, a China-state-aligned actor used ARTEX to attempt authentication against more than 2,250 downstream Microsoft 365 tenants.

According to the FBI, ARTEX "enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously, fundamentally accelerating attack velocity and expanding operational scale." The shift matters more than any single IOC: defenders now face less-skilled operators working at machine speed against many targets at once.

What ARTEX Is

ARTEX is an autonomous penetration testing framework written in Go. It uses AI agents to handle reconnaissance, vulnerability discovery, exploitation, privilege escalation and data theft with little human input. It does not ship its own model. It connects to external large language models, including Anthropic Claude, OpenAI GPT, DeepSeek, GLM, Grok and Kimi.

The operator states goals and tasks. ARTEX breaks them into commands and hands them to specialized agents that carry them out. That design lets one operator run intrusion activity against several organizations at once. The FBI says the framework is especially good at finding weaknesses in systems defenders tend to overlook: partner portals, employee-facing applications, legacy services and third-party integrations.

โš ๏ธ Takedown did not stop distribution โ€” Developer Autumn-27 published ARTEX on GitHub in July 2026 and deleted the repository on October 8 after reports of abuse. The FBI says the framework still circulates through community forks, archived copies, localized versions and a commercial edition sold as ARTEX PRO. Researchers counted more than 1,600 publicly exposed ARTEX consoles as of early November, and the real number of deployments is believed to be much higher.

The ConnectWise Supply-Chain Attack

The most significant ARTEX-linked campaign in the advisory is the October compromise of ConnectWise Cloud Backup, formerly SkyKick Cloud Backup. The attackers compromised the application identities the backup service uses to reach customer Microsoft 365 environments. Customers had already granted those identities permissions, so the attackers could authenticate to customer tenants as a trusted application. They never had to phish a single user.

Microsoft attributed the campaign to Storm-0940, a China-state-aligned actor. Its October investigation found that Storm-0940 used ARTEX to speed up reconnaissance, password spraying and abuse of cloud application permissions. After taking over the ConnectWise identities, the group used ARTEX to attempt authentication against more than 2,250 downstream tenants. Some intrusions progressed to credential modifications and attempts to establish persistence. Huntress reported at least 162 compromised tenants at the time.

This is the pattern to watch for. A single trusted third-party integration gave access to thousands of environments, and AI automation let the attackers work through that access far faster than a human team could.

South Korean Financial Sector Campaign

The FBI also cites attacks on South Korean financial institutions that exposed personal information on about 67,000 people. In that campaign, a Chinese-speaking threat actor paired ARTEX with Claude Code agents. Taken together, the two campaigns show ARTEX being used by both state-aligned groups running supply-chain operations and by financially motivated actors going after regulated data.

Why Speed Is the Core Problem

Automated attacks are noisy. ARTEX generates large volumes of suspicious traffic, and that traffic often triggers alerts. The problem is that detection alone does not help much when the whole chain finishes inside a single analyst shift. If privileged access arrives in under three hours, an alert sitting in a queue overnight is an alert that arrived after the breach.

The FBI's guidance focuses on the points where ARTEX does the most damage:

  • Securing internet-facing applications, including partner and employee portals that often sit outside the main hardening program
  • Restricting access to cloud administration interfaces
  • Reviewing application permissions and authentication logs, particularly for third-party apps with standing tenant access
  • Rate limiting sensitive APIs to slow automated enumeration and spraying
  • Monitoring for bursts of automated requests, unusual cloud activity, unexpected credential or privilege changes, and large outbound data transfers

Detection Resources and Their Limits

The 13-page FBI advisory includes indicators of compromise from known ARTEX campaigns, including IP addresses, command-and-control infrastructure and suspicious network activity. The bureau warns that the listed IPs may have changed since the attacks took place.

On the same day, researcher MalwareMustDie released an open-source ARTEX detection signature pack. It covers suspicious network requests, identity abuse, cloud activity, malicious files and automated attack patterns across several monitoring and detection platforms. The accompanying playbook sorts indicators by confidence level. It notes that IP addresses and other infrastructure indicators go stale quickly and can produce false positives.

Behavior outlasts infrastructure โ€” ARTEX runs on commodity LLM APIs and operator-chosen hosting, so IP-based blocking has a short shelf life. Behavioral signals hold up longer: request bursts, identity abuse by application principals, rapid privilege changes and bulk data movement. Organizations that find ARTEX-related activity are encouraged to report it to the FBI or IC3.

RedEye Assessment

ARTEX is not a new exploit. What it changes is cost. Reconnaissance and attack chaining that used to take a skilled team days now runs unattended across many targets in hours. The ConnectWise case shows where that hurts most: third-party application identities with broad, rarely reviewed permissions into Microsoft 365. Expect more forks, more commercial variants and more actors adopting ARTEX now that the code cannot be pulled back. Organizations should find out which integrations hold standing access to their tenants and whether their detection-to-response time fits inside a three-hour window. RedEye can assess both.

Originally published on RedEye Threat Intelligence.

๐Ÿ“ฐ Read the original article on Dev.to Security

Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes โ€” full credit and traffic to the original publisher.