Attackers used a forked agentic pentest tool to reach AWS production secrets, Arctic Wolf says
Arctic Wolf links self-hosted copies of AntiFox, an open-source agentic pentest tool, to intrusions that reached AWS production secrets.
Security firm Arctic Wolf has linked self-hosted copies of AntiFox, an open-source agentic penetration-testing platform, to several intrusions against web servers and Amazon Web Services accounts. In one case the attackers obtained a production secret. Arctic Wolf published its findings on 10 November, and BleepingComputer reported them on 11 November.
What AntiFox is
According to BleepingComputer, AntiFox is a fork of version 0.1.1 of ARTEX, another agentic pentesting tool. It has a Chinese-language interface and lets users deploy AI agents that automate reconnaissance tasks such as mapping exposed services and probing endpoints for weaknesses. Its publishers prohibit use against live systems or for illegal activity. BleepingComputer notes that they describe no technical controls to enforce this.
What Arctic Wolf observed
In its technical report, Arctic Wolf says Alibaba Cloud systems hosting AntiFox interfaces were associated with confirmed malicious actions in "several recently observed intrusions". It does not give an exact count. The activity included:
- A web server compromise. On an internet-facing IIS server, the attackers created an ASPX web shell, ran system commands and read configuration files containing database connection strings and application settings. They then copied configuration and log files into a web-accessible upload folder. Arctic Wolf says it did not identify the vulnerability used to get in.
- Rapid AWS enumeration. Using valid IAM users, and in some cases an assumed EC2 role, the attackers made hundreds of API requests across dozens of AWS services in about 30 minutes, according to BleepingComputer. Most were denied, but some succeeded. One chain ended with a successful request for a production secret.
- S3 activity. In a separate case, a valid AWS identity was used to read, copy and delete objects in S3 storage.
Arctic Wolf describes the link to AntiFox as an association with infrastructure, not a firm attribution. It names no threat group and notes that public cloud addresses can be reassigned or shared with unrelated tenants.
Amazon's response
In an update to the BleepingComputer story, an Amazon spokesperson said AWS services "were not compromised and operated as designed". The spokesperson said the research describes attackers using credentials taken from customers' compromised applications. Amazon recommended least-privilege permissions, requiring IMDSv2, keeping long-term access keys out of application configuration files, and turning on Amazon GuardDuty.
Why it matters
As the report describes it, the intrusion techniques themselves are familiar: web shells, credentials harvested from configuration files and cloud enumeration. What has changed is the tooling. Agentic frameworks can go through many services and credentials far faster than a person working by hand. Arctic Wolf's own observation was a burst of mostly denied API calls with successful ones scattered among them.
BleepingComputer also notes that CrowdStrike has previously linked the parent tool, ARTEX, to a Chinese group it calls Vapor Panda, which it says targeted more than 200 organisations across 14 countries. Those cases are separate from the AntiFox intrusions, and Arctic Wolf does not connect them to a named group.
What defenders are advised to do
Arctic Wolf's recommendations, as summarised by BleepingComputer, include:
- correlating IIS web shell activity with unusual AWS access
- reviewing successful API calls that appear within bursts of denied calls
- revoking affected AWS sessions and keys
- rotating any credentials exposed through application files.
The report also lists indicators, including two Alibaba Cloud IP addresses, AntiFox web interface artefacts on TCP port 8787, and characteristic AWS API sequences such as ListSecrets followed by GetSecretValue. Arctic Wolf says the list is not exhaustive.
Sources
- BleepingComputer: Hackers use AntiFox AI agents in AWS credential-theft attacks
- Arctic Wolf Labs: AntiFox report on GitHub