FBI warns criminals use ARTEX AI agents to speed intrusions
Original title: Criminals are using ARTEX AI agents to speed up cyberattacks, FBI warns
Some observed attacks reached privileged access in under three hours, with abuse reported since at least August.
An FBI FLASH advisory on November 13 says attackers are abusing ARTEX, an open-source penetration-testing framework whose planning agent assigns tasks to other agents that call external large language models. The bureau says use has been seen since at least August against finance, government, healthcare, and technology organizations worldwide. Separately, Microsoft reported that a China-state-aligned group used ARTEX after compromising ConnectWise application credentials. Recommended defenses include patching exposed systems, rotating secrets, and watching for rapid activity.
Unconfirmed: The China-state-aligned attribution is Microsoft's separate report and does not name the group.
- ARTEX has no model of its own and connects to external LLMs.
- Once a task begins, the framework needs relatively little human input.
- Microsoft linked one case to stolen ConnectWise application credentials.
- FBI guidance covers internet-facing apps, admin paths, and secret rotation.
Impact: Automated intrusions that can reach privileged access
- 1.Patch internet-facing systems and inventory old employee and supplier portals.
- 2.Scan for exposed passwords and API keys, then rotate compromised secrets.
- 3.Watch for rapid request bursts and unexpected credential or privilege changes.
- ATT&CK
- T1078 Valid Accounts
- Vendors
- ConnectWise
- Products
- ARTEX
- Countries
- China
From the original article
2 min read at malwarebytes.com
The FBI warns that attackers are abusing an AI security-testing tool, with some intrusions reaching privileged access in three hours.
By @MetallicaMVP · malwarebytes.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.