AntiFox platform linked to attacks stealing AWS production secrets

Cybersecurity researchers have disclosed a campaign in which an AI-driven autonomous pentesting platform named AntiFox has been used to steal AWS credentials and pivot into production environments. Arctic Wolf Labs described activity linked to Alibaba Cloud hosting AntiFox interfaces, with intrusions involving web exploitation and web shell deployment to enable configuration discovery, credential validation, cloud-service enumeration, secret access, and storage operations. The tool is downloaded from GitHub, and the project had been forked 251 times and starred 474 at the time of reporting.

Initial access included a compromised internet-facing IIS application, enabling the attacker to create an ASPX web shell using the IIS worker process (w3wp.exe) and execute OS commands. Actions observed included locating web.config files and backups, reading application settings and database connection strings, enumerating log directories, attempting to reach an internal production database with credentials from the application config, and copying configuration and log files to a web‑accessible upload directory.

Arctic Wolf noted that the use of w3wp.exe to run commands and the subsequent web shell suggests compromise via the affected web application. AntiFox’s documentation describes a multi-agent system performing recon, attack-surface discovery, exploitation, finding capture, and report delivery, emphasising delivery over discovery.

The analysis also notes that some initial IAM-based access attempts to AWS services (IAM, EC2, S3, Secrets Manager, etc.) ended with AccessDenied, though some GetSecretValue calls and S3 operations were successful in certain instances.