Kivorent News

Wednesday, November 18, 2026

Tech and AINews

FBI warns attackers are using the ARTEX AI hacking framework

The FBI has issued a FLASH advisory warning that multiple threat actors have used ARTEX, an open-source tool that strings AI agents together to break into networks, since at least August 2026.

Warden, covering security, outages and the rules for tech
, 2 min read

<3 hoursfastest intrusions from first targeting to privileged access

What to know

  • The FBI's FLASH advisory, released November 13, says multiple threat actors have used ARTEX since at least August 2026 against financial, government, technology, defense and healthcare targets.
  • ARTEX is a Go framework whose AI agents handle reconnaissance, exploitation, privilege escalation and data theft, using outside models such as Claude, GPT, DeepSeek, GLM, Grok and Kimi.
  • There is no patch because ARTEX is a tool, not a flaw; the FBI advises locking down internet-facing apps and cloud admin access and watching for bursts of automated activity.

The FBI has warned that multiple threat actors are using ARTEX, an AI-driven penetration testing framework, to break into organizations around the world. The bureau released a 13-page FLASH advisory on Friday, November 13, according to BleepingComputer, which reviewed it. The advisory says attackers have used the tool since at least August 2026 against the financial, government, technology, defense, healthcare and other sectors, with some intrusions going from first targeting to privileged access in under three hours.

Kivorent News could not open the FBI's advisory itself, so the details below come from BleepingComputer's report and an earlier analysis by threat intelligence firm SOCRadar.

What ARTEX is

ARTEX is written in Go and uses coordinated AI agents to run the stages of an attack: reconnaissance, finding and exploiting weaknesses, escalating privileges and stealing data. It does not ship its own model. Instead it connects to outside large language models, and BleepingComputer lists Anthropic's Claude, OpenAI's GPT, DeepSeek, GLM, Grok and Kimi among them.

The tool appeared on GitHub in July 2026, presented as a framework for learning, research and testing systems the user is authorized to test, SOCRadar says. Its developer stopped open-source updates on October 8, citing misuse, but forks remain public, and BleepingComputer reports a paid edition called ARTEX PRO. BleepingComputer says more than 1,600 ARTEX consoles were exposed on the internet as of early November.

Where it has turned up

BleepingComputer ties ARTEX to several campaigns this fall:

  • ConnectWise Cloud Backup (formerly SkyKick Cloud Backup): in October, Huntress counted at least 162 compromised tenants. Microsoft attributed the campaign to Storm-0940, a group it calls China-state-aligned, and says attempts reached more than 2,250 downstream Microsoft 365 tenants.
  • South Korean banks: personal data of about 67,000 people was taken from Shinhan Bank, KB Kookmin Bank and Hana Bank. South Korean police separately arrested seven suspects linked to attacks on more than 90 organizations in which about 8.2 million records were stolen.
  • Vapor Panda: a China-linked espionage group, as CrowdStrike calls it, targeted more than 200 organizations in 14 countries and compromised at least nine.

What the FBI recommends

Because ARTEX is a tool rather than a vulnerability, there is no CVE and nothing to patch. The FBI advises securing internet-facing applications, restricting access to cloud administration interfaces, reviewing application permissions and authentication logs, and rate-limiting sensitive APIs. It also says to watch for bursts of automated requests, unexpected credential or privilege changes and large data transfers. The bureau cautions that attacker IP addresses listed in the advisory may already have changed.

Why it matters

ARTEX lets attackers hand much of an intrusion to AI agents, which shortens the time defenders have to notice. The fixes are ordinary ones, such as tight cloud permissions, rotated credentials and quick patching of known exploited flaws, but they now have to hold against attacks measured in hours. Stolen cloud and AI credentials were also central to Railway's October breach and Cursor's marketplace hack. For background, see how to read a security advisory.

Sources

AI agentsAI modelsInternet, outages and security

Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.