TLDR
- The FBI says some ARTEX-driven intrusions reached privileged access in under three hours.
- ARTEX is an open-source security testing framework that lets AI agents run attack steps with little human input.
- Attackers tried to reach more than 2,250 Microsoft 365 tenants, but that figure measures targeting, not confirmed breaches.
- One agent went from a leaked GitHub secret to Azure Key Vault data in five minutes.
- More than 1,600 ARTEX consoles were exposed online by early November, though exposure alone does not prove malicious use.
The FBI says hackers are using an AI tool called ARTEX to break into networks in as little as three hours. Its alert, dated November 13, 2026, describes attacks that reached privileged access that fast.
Threat actors are weaponizing ARTEX, an open-source, large language model (LLM)-driven agentic platform to automate multiple phases of cyber intrusions at machine speed. Since at least August 2026, a variety of malicious cyber actors have used ARTEX to conduct campaigns targeting… pic.twitter.com/Spvq87XzsR
— FBI Cyber Branch (@FBICyberBranch) November 13, 2026
The three-hour figure describes some observed intrusions. It is not a fixed speed for every attack.
ARTEX is an open-source security testing framework, not a malware family. It is written in Go and connects outside large language models to software agents.
Operators set a goal, and the agents scan targets and run tools. They also review results and retry failed steps with little human input.
Researchers say the tool links known weaknesses instead of relying on one new exploit. Those weaknesses include stolen credentials, exposed applications and loose cloud permissions.
How the Attacks Unfold
The alert lists SQL injection, missing API authentication and exposed passwords. It also names the abuse of trusted cloud connections.
Once inside, agents added credentials to existing applications and deployed web shells. They also tried to steal credentials and searched for Windows systems without endpoint protection.
One agent mapped more than 110,000 directory objects. Google saw single automated bursts last only two to 25 seconds.
Microsoft linked ARTEX to Storm-0940, a China-aligned group that targeted Microsoft 365 environments. The group used compromised ConnectWise application credentials.
Attackers tried to reach more than 2,250 tenants using permissions customers had already approved. That number shows targeting, not confirmed breaches.
A separate investigation found agents searching GitHub for exposed secrets. In one case, an agent went from repository searches to stolen Azure Key Vault data in five minutes.
A group tracked as VAPOR PANDA targeted organizations in 14 countries, including foreign ministries and defense contractors. Researchers confirmed post-exploitation at nine of more than 200 targets.
Criminal campaigns recovered over 300 code-signing certificates and 190,000 gaming credentials. Model costs ran from $145 to $1,500 per operation.
What the Evidence Does Not Show
WorkOS found unauthorized access to a sales application holding customer contact details and Slack excerpts. Its CEO named ARTEX as involved.
The company reported no evidence that production services or WorkOS application data were affected.
Researchers also examined servers exposed by an operator tied to South Korean financial attacks. Those attacks affected about 67,000 people.
The files showed AI routing tools, reusable instructions, proxy services and stored credentials. Shared use of ARTEX does not prove that this operator, Storm-0940 and the WorkOS attacker are connected.
Publicly exposed ARTEX consoles exceeded 1,600 by early November. Exposed installations do not automatically prove malicious use by their operators.
