AntiFox AI Pentesting Platform Targets AWS Production Secrets in Active Cloud Intrusions
November 11, 2026
Cybersecurity researchers at Arctic Wolf Labs have uncovered an active campaign where attackers abuse AntiFox, an open-source AI-driven penetration testing platform, to steal Amazon Web Services (AWS) credentials and access production secrets. The findings highlight a growing threat in which agentic AI frameworks are weaponized to automate cloud intrusions at machine speed.
What Happened: AntiFox AI Agents Breach AWS Environments
Arctic Wolf Labs disclosed that Alibaba Cloud systems hosting AntiFox interfaces were tied to confirmed malicious actions against internet-facing web services and associated AWS environments. In several intrusions, attackers deployed an ASPX web shell on compromised IIS servers. Consequently, they executed OS commands and harvested application settings and database connection strings.
The attacker then used valid AWS identities, including IAM users and assumed EC2 roles, to make hundreds of authenticated API requests in about 30 minutes. Moreover, some of these requests succeeded, granting access to production secrets stored in AWS Secrets Manager and enabling read, copy, and delete operations on S3 objects.
Technical Details of the AntiFox Cloud Intrusion
AntiFox is an AI-driven autonomous penetration testing platform forked from ARTEX 0.1.1. It runs a planner / worker / main-agent multi-agent system over a dual-graph core. Furthermore, it automates reconnaissance, attack-surface discovery, exploitation, and reporting.
In the observed attacks, the threat actors performed the following actions:
- Deployed an ASPX web shell via the IIS worker process
w3wp.exe - Ran PowerShell commands to locate
web.configfiles and backups - Read application settings and database connection strings
- Enumerated debug and error-log directories
- Copied configuration and log files into a web-accessible upload directory
- Used valid IAM credentials to validate access and enumerate AWS services
- Made hundreds of cross-service API calls across IAM, EC2, S3, Lambda, Secrets Manager, KMS, RDS, ECS, EKS, and others
While many requests were blocked with “AccessDenied” errors, some discovery and secret-access operations succeeded. Therefore, the attackers obtained production credentials in at least some instances.
Business and Operational Impact
The abuse of agentic AI platforms like AntiFox lowers the barrier to entry for sophisticated cloud attacks. The operational impacts include:
- Credential exposure: Valid AWS IAM credentials and EC2 roles were compromised, allowing persistent cloud access.
- Data theft: Attackers successfully accessed production secrets and performed S3 object operations.
- Lateral movement: Web shell deployment on IIS enabled deeper network reconnaissance and data staging.
- Speed and scale: Hundreds of API requests in 30 minutes demonstrate machine-speed automation that outpaces human defenders.
- Supply-chain risk: AntiFox remains publicly available on GitHub without technical controls to prevent misuse.
Mitigation and Recommendations
Immediate Actions for Defenders
- Correlate IIS web shell activity with unusual AWS API access patterns.
- Review successful API calls within bursts of denied requests.
- Revoke affected AWS sessions and rotate exposed credentials.
- Audit application configuration files for hardcoded secrets.
- Restrict IAM permissions using least-privilege policies.
Long-Term Security Measures
Organizations should also consider cloud security posture management tools and behavior analytics to detect agentic attack patterns. Furthermore, enabling AWS CloudTrail logging and GuardDuty anomaly detection will help identify bursts of cross-service API calls. In addition, security teams should monitor for Alibaba Cloud IP ranges if they are not part of the expected traffic profile.
Bottom line: The AntiFox campaign shows that open-source AI pentesting tools are already being abused in the wild. Defenders must treat agentic frameworks as an emerging threat vector and harden both on-premise web applications and cloud identity controls accordingly.
Incident Summary
| Incident: | AntiFox AI platform abused to steal AWS production secrets |
| Affected Systems: | Internet-facing IIS servers, AWS IAM/EC2/S3/Secrets Manager environments |
| Disclosure Date: | November 11, 2026 |
| Patch Status: | N/A — mitigation focuses on IAM hardening, credential rotation, and web shell detection |
| Threat Actor: | Unknown; infrastructure hosted on Alibaba Cloud |
| Reporter: | Arctic Wolf Labs |
References
- Arctic Wolf Labs, “AntiFox: Alibaba-Hosted Platforms Facilitate Web and Cloud Intrusions,” November 2026, https://github.com/rtkwlf/wolf-tools/tree/main/pack_alerts/202611-antifox-alibaba-hosted-platforms-facilitate-web-cloud-intrusions, accessed November 11, 2026.
- Bill Toulas, “Hackers use AntiFox AI agents in AWS credential-theft attacks,” BleepingComputer, November 11, 2026, https://www.bleepingcomputer.com/news/security/hackers-use-antifox-ai-agents-in-aws-credential-theft-attacks/, accessed November 11, 2026.
- Arctic Wolf Labs, “AntiFox AI Pentesting Platform Linked to Attacks That Accessed AWS Production Secrets,” The Hacker News, November 11, 2026, https://thehackernews.com/2026/11/antifox-ai-pentesting-platform-linked.html, accessed November 11, 2026.
- AntiFox GitHub Repository, https://github.com/AntiFox-AI/antifox, accessed November 11, 2026.