CyberStrikeAI 16-Day Automated AWS Intrusion Analysis
Score: 9/10

CyberStrikeAI 16-Day Automated AWS Intrusion Analysis

A CyberStrikeAI server targeted four AWS accounts over 16 days, leveraging machine-speed API calls to harvest secrets and execute remote commands via SSM.

Executive Summary

Between September 30 and October 16, 2026, an adversary utilizing the CyberStrikeAI platform—an open-source framework pairing LLMs with offensive tools—conducted a multi-phase intrusion into four AWS accounts. The attack originated from the IP 43.240.29[.]11 and demonstrated significant automation, characterized by a burst of 56 EC2 user data requests in just 141 milliseconds.

The attack chain progressed from credential validation (STS GetCallerIdentity) to harvesting Secrets Manager values and EC2 startup scripts (userData), which often contain hardcoded credentials. Most notably, the actor decrypted SSM SecureString parameters and subsequently issued SSM SendCommand executions, potentially achieving remote code execution on target instances. The campaign was notable for a 14-day operational gap, designed to bypass short correlation windows in traditional monitoring tools.

This activity highlights the emergence of agentic offensive platforms that can operate at machine speeds while maintaining a low detection profile. The high volume of automated medium-severity alerts underscores the need for IP-based correlation and robust cloud identity protection strategies.

Key Details

Threat Name

CyberStrikeAI

Affects

—

Adversary

—

Malware/Tools

CyberStrikeAI

Report Score

9out of 10
Quality Score
Excellent
IOC Quality9
TTP Details9
Detection Guidance8
Enterprise Relevance10
Clarity & Structure10
Technical Depth7

Sources