The FBI has warned that multiple hacking groups are using an AI-powered penetration-testing tool called ARTEX to automate cyberattacks on organisations worldwide, in some cases going from first targeting to privileged access in under three hours.
In a FLASH advisory, the bureau said attackers have used the open-source framework since at least August 2026 against financial, government, technology, defence and healthcare organisations.
“The framework enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously, fundamentally accelerating attack velocity and expanding operational scale,” the FBI said.
What ARTEX does
ARTEX uses AI agents to carry out reconnaissance, find and exploit vulnerabilities and steal data with little human input. It doesn’t have its own AI model; it connects to external ones, including Anthropic’s Claude, OpenAI’s GPT, DeepSeek, GLM, xAI’s Grok and Kimi.
The FBI says it is especially good at finding weaknesses in overlooked systems — partner portals, employee apps, legacy services and third-party integrations.
Its developer removed it from GitHub on 8 October after reports of abuse, but copies, forks and a commercial “PRO” edition remain available. Researchers had found more than 1,600 publicly exposed ARTEX consoles by early November.
Known attacks
The advisory links ARTEX to several incidents:
- ConnectWise supply-chain attack (October): Microsoft attributed it to China-aligned group Storm-0940, which used ARTEX to attempt access to more than 2,250 downstream Microsoft 365 customer environments; Huntress identified at least 162 compromised.
- South Korean banks: a Chinese-speaking attacker combined ARTEX with Claude Code agents against Shinhan, KB Kookmin and Hana banks, exposing personal data of about 67,000 people.
- South Korean gang: police arrested seven suspects over attacks on more than 90 organisations, stealing about 8.2 million records, mostly using ARTEX.
- Vapor Panda: CrowdStrike says this Chinese espionage group used ARTEX against more than 200 organisations in 14 countries, compromising at least nine.
What defenders should do
The FBI urges organisations to secure internet-facing applications, restrict access to cloud admin interfaces, review app permissions and login logs, and rate-limit sensitive APIs. It also recommends watching for bursts of automated requests, unexpected credential or privilege changes and large data transfers. Its advisory includes indicators of compromise, and researcher MalwareMustDie has released free detection rules.





