Malicious actors have begun weaponizing AntiFox, an open-source agentic penetration testing platform, by hosting instances on Alibaba Cloud to conduct attacks against internet-facing web services and AWS infrastructure. The compromised infrastructure has facilitated multiple attack phases, ranging from vulnerability discovery and webshell implantation to credential theft and harvesting of AWS secrets.

AntiFox originated as a derivative of ARTEX 0.1.1 and features a Chinese-language interface designed to streamline penetration testing workflows. The platform enables operators to deploy autonomous AI agents capable of automating reconnaissance activities, including service enumeration and endpoint vulnerability scanning.

image

Researchers at Arctic Wolf identified malicious AntiFox deployments on Alibaba Cloud infrastructure that leveraged legitimate AWS identities—specifically IAM users and assumed EC2 roles—to authenticate API requests against Amazon Web Services. According to their findings, "In several recently observed intrusions, Alibaba Cloud systems hosting AntiFox interfaces were associated with confirmed malicious actions against internet-facing web services and associated AWS environments."

Attack Patterns and Tactics

One documented intrusion involved deploying an ASPX webshell to an exposed IIS server, enabling remote command execution and access to application configuration data and database connection strings. The attacker subsequently exfiltrated configuration files and logs to a web-accessible location.

In the same incident, the threat actor utilized valid IAM credentials and EC2 roles to probe AWS API endpoints. Within a 30-minute window, the attacker cycled through multiple identities and issued hundreds of API requests. While the majority were denied, some succeeded, resulting in the extraction of production secrets. A separate case demonstrated the attacker using valid AWS credentials to read, modify, and remove S3 objects.

Defensive Recommendations

Arctic Wolf's analysis provides defenders with actionable guidance for detection and response:

Broader Context

The AntiFox campaign mirrors earlier activity involving ARTEX AI, a competing agentic penetration testing framework developed in China. ARTEX transitioned to closed-source distribution following its use in attacks targeting South Korean financial institutions. CrowdStrike attributed ARTEX exploitation to Vapor Panda, a Chinese-nexus threat group that deployed the framework during a cyberespionage operation spanning more than 200 targets across 14 countries.

AntiFox continues to be distributed via GitHub and has reached version 2.0, which enhances documentation and reporting capabilities. The project's terms of service explicitly prohibit deployment against production systems, real-world penetration testing engagements, security exercises, and destructive or unlawful purposes. However, the platform lacks built-in technical safeguards to prevent such misuse.