Skip to content
Wed 18 Nov 2026 Toronto Sign in
Privacy & Security

AntiFox: attackers are using an AI pentest agent to steal AWS secrets. Six checks to run

Arctic Wolf says attackers ran self-hosted AntiFox, an ARTEX fork, to move from IIS webshells to AWS secrets. Here's how it worked and what to check.

Abstract illustration: a chain of five outlined boxes linked by lines on a dark red background
Illustration: prompt/power

A free, Chinese-language “AI pentesting” platform is turning up in real intrusions, and the target is the cloud account sitting behind ordinary Windows web servers. Security firm Arctic Wolf says attackers have been running self-hosted copies of AntiFox, an agentic penetration-testing tool, on Alibaba Cloud to break into web servers and then harvest Amazon Web Services secrets.

Here’s what was reported, and the checks worth running this week if your company runs IIS servers that talk to AWS.

What AntiFox is

AntiFox is an open-source platform that lets users deploy AI agents to map exposed services and probe them for vulnerabilities. It began as a fork of ARTEX 0.1.1, the China-developed agentic pentesting framework that went closed-source after it was used in attacks on South Korean banks, and is now at version 2.0 on GitHub, according to BleepingComputer’s Nov. 11 report on Arctic Wolf’s research. Its publishers forbid use against live systems or for illegal activity, but nothing in the software enforces that.

Five-step diagram: exposed IIS server with ASPX webshell, config and connection strings read, AWS IAM users and EC2 roles used, burst of hundreds of API calls in about 30 minutes, production secrets taken and S3 objects copied and deleted
The attack chain Arctic Wolf described. Source: Arctic Wolf via BleepingComputer. Graphic: prompt/power

How the intrusions worked

In the cases Arctic Wolf described, the attacker had an ASPX webshell on an internet-facing IIS server. The report doesn’t say how that first foothold was gained. From there, the attacker ran operating-system commands, read application settings and database connection strings, and copied configuration and log files into a web-accessible folder.

The next step is the one to worry about. Using valid AWS identities, including IAM users and in some cases assumed EC2 roles, the attacker made authenticated AWS API calls. In one intrusion they cycled through several identities and fired hundreds of requests in about 30 minutes. Most were denied; some weren’t, and production secrets were taken. In a separate incident, a valid identity was used to read, copy and delete S3 objects.

Amazon told BleepingComputer that “AWS services were not compromised and operated as designed,” and that the activity relied on credentials taken from customers’ compromised applications.

Six checks to run now

These come from Arctic Wolf’s guidance and Amazon’s recommendations as reported. None needs new software.

  1. Hunt for webshells on IIS. Look for unexpected .aspx files in web roots and for config or log files copied into public folders.
  2. Correlate the two halves. Line up suspicious IIS activity with unusual AWS API access from the same period; that pairing is the signature here.
  3. Read the successes inside the failures. In CloudTrail, find bursts of AccessDenied errors and check which calls in that burst succeeded.
  4. Revoke and rotate. Kill active sessions and keys for any identity involved, and rotate every credential stored in application files on an affected server.
  5. Get long-term keys out of config files. Amazon recommends keeping long-term access keys out of application configuration and granting least privilege.
  6. Require IMDSv2 and turn on GuardDuty. Enforcing IMDSv2 on EC2 instances makes role credentials harder to steal; GuardDuty flags anomalous API use.

The bigger pattern

Agentic attack tools don’t need to be clever to be dangerous. They make the boring parts of an intrusion fast: trying many identities, many API calls, many minutes in a row. That is why the defensive advice above is about speed of detection and about removing stored secrets, not about AI. Arctic Wolf’s full report includes hunting queries for security teams.

More from prompt/power: Check your coding extensions for a backdoor · Cloudflare Workers AI data leak · How to lock down AI agents

// How we reported this

Researched and drafted with AI tools, fact-checked against primary sources and approved by a human editor before publishing. Prices and dates are as of publication. Our standards →

Columnist, Consumer Tech Lauren Smith

Lauren Smith covers consumer apps and services for prompt/power: streaming, subscriptions, e-commerce and everyday tech worth your money. She keeps a running tally of every free trial that quietly turned into a subscription.

All stories by Lauren →

Latest from prompt/power

  1. Nvidia beats its own forecast with US$115.8 billion, then guides margins lowerNov 18
  2. Apple, the NSA and Chase: what each says broke in a week of outagesNov 18
  3. ChatGPT for Healthcare now has agents that click through hospital software. The fine print is all AmericanNov 17
  4. ChatGPT Wallet lets an AI agent pay for you. Here’s who can use it, and why Canada and the U.K. waitNov 17
  5. Gemini 4 Argon’s API price doubles Dec. 1. Here’s what it costs in US$, CA$ and £Nov 17

Leave a Reply

Your email address will not be published. Required fields are marked *