AI News

Open-Source AI Penetration Testing Framework ARTEX Draws FBI Attention: Attackers Gain Highest Privileges in Under 3 Hours

cnBetaUpdated Nov 16 21:03

AI summary

An FBI bulletin states that the open-source AI multi-agent penetration testing framework ARTEX has been used in attacks, taking less than 3 hours from locking onto a target to gaining the highest privileges. Google's threat intelligence team says that in actual incident response it observed attackers using ARTEX for post-compromise reconnaissance, privilege escalation, and lateral movement, with some cases first providing leaked credentials and then letting the framework carry out subsequent operations. Related attacks include a campaign against South Korean financial institutions that resulted in the leakage of personal information of about 67,000 people; the original ARTEX repository has been shut down, but derivative versions are still circulating, and the FBI recommends that enterprises investigate leaked keys and monitor for new credentials and anomalous activity in high-privilege applications.

Why read this

AI score 76

An FBI bulletin discloses that ARTEX took less than 3 hours from locking onto a target to gaining the highest privileges, and provides investigation directions for credentials and cloud activity.

Topics

Tags

Body · AI translation

ARTEX is an AI-based multi-agent penetration testing platform that can convert natural-language descriptions of targets into attack calls and operational steps. The FBI says this lowers the technical barrier required for organizations to conduct multi-target attacks and also shortens the time needed to chain attack steps together.

Google's threat intelligence team also stated that security researchers have observed attackers using ARTEX for post-compromise reconnaissance, privilege escalation, and lateral movement during actual incident response. In some cases, attackers first provide the framework with already-leaked account credentials, then have the framework carry out subsequent operations.

The FBI bulletin also mentioned that the related attacks include a campaign against South Korean financial institutions that resulted in the leakage of personal information of approximately 67,000 people; at the time, multiple South Korean financial institutions were targeted, and in addition to the information leak, the attacks also caused massive economic losses.

The original ARTEX repository has been closed, but derivative versions are still circulating:

The original author of the ARTEX penetration testing framework is a threat detection engineer at the domestic security company ThreatBook, and this project was a side project of this threat detection engineer. After someone used the framework to attack South Korean financial institutions in October, the project author closed the original repository and announced that it would no longer be open source.

Earlier, ThreatBook also issued a statement saying that it owns the copyright to the ARTEX framework, and that the previously open-sourced version will continue to be provided under the original license. There are currently many derivative versions based on ARTEX online, and these derivative versions are being used by more hackers to launch automated attacks.

The FBI recommends that enterprises check credentials and cloud activity:

The FBI recommends that organizations focus on protecting internet-exposed applications, cloud infrastructure, and partner systems, and conduct investigations using the indicators of compromise (IOCs) in the bulletin. The specific measures are as follows:

Monitor for anomalous activity involving newly added credentials, permission assignments, and high-privilege applications

Check repository code, configuration files, environment variables, and CI/CD pipelines for leaked secrets, and promptly rotate exposed credentials

Implement server-side authorization and access rate limiting for partner and employee portals, and monitor for anomalous bulk queries

Retain identity, audit, and security monitoring logs, and after confirming anomalies, isolate and remove unauthorized access

Source

cnBeta