Summary：**Security Experts Warn: AI‑Generated PowerShell Threatens Active Directory**Cybersecurity researche

referrerpolicy="no-referrer"
style="max-width:100%;height:auto;display:block;margin:0 auto;">

**Security Experts Warn: AI‑Generated PowerShell Threatens Active Directory**

Cybersecurity researchers have uncovered a new intrusion in which an unknown threat actor used an AI‑crafted PowerShell script to enumerate Active Directory (AD) environments. The script, described by analysts as “vibe‑coded,” automatically located the domain controller, then harvested lists of users, computers, and nested domains before exfiltrating the data to an external server. While the attack did not modify AD objects, the reconnaissance phase gives adversaries a detailed map for later privilege‑escalation or ransomware campaigns.

**Key Developments**
The discovery came from a forensic review of a compromised Windows server belonging to a mid‑size financial services firm. Investigators noted that the PowerShell payload lacked typical obfuscation markers; instead, it displayed natural‑language comments and variable names that resembled output from a large‑language model. The script executed via a scheduled task that triggered after a user opened a malicious email attachment, illustrating how attackers are blending social engineering with generative‑AI tools to bypass traditional signature‑based defenses. Researchers shared indicators of compromise (IOCs) including specific registry keys, PowerShell command lines, and the command‑and‑control (C2) domain used for data upload.

**Industry Analysis**
Experts say the incident signals a shift in the threat landscape: adversaries are now leveraging AI to accelerate the creation of custom reconnaissance tools, reducing the time and expertise needed to build effective AD‑enumeration scripts. “What used to take a skilled red‑team operator hours to develop can now be generated in minutes with a prompt to an LLM,” said one senior analyst at a threat‑intelligence firm. This lowers the barrier for less‑technical actors and increases the volume of targeted probes against enterprise directories. Defensive teams must therefore update detection rules to