<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>preemware</title><link>https://preemware.io/</link><description>Offensive security and LLM red teaming.</description><generator>Hugo 0.164.0</generator><language>en-US</language><atom:link href="https://preemware.io/index.xml" rel="self" type="application/rss+xml"/><lastBuildDate>Sun, 26 Jul 2026 12:00:00 -0500</lastBuildDate><item><title>AI Pentesting System Roots Active Hard Hack The Box Machines</title><link>https://preemware.io/posts/ai-pentesting-active-hard-hack-the-box-machines/</link><pubDate>Sun, 26 Jul 2026 12:00:00 -0500</pubDate><guid>https://preemware.io/posts/ai-pentesting-active-hard-hack-the-box-machines/</guid><description>Most autonomous-pentesting results have a contamination problem.
When a target has been retired and its walkthroughs have been public for years, it is difficult to know whether a model reasoned through the machine or reconstructed a path already present in its training data or retrieval context. Evaluating against active targets reduces that obvious source of contamination. It does not prove that a model had zero prior exposure, but it makes the result more informative.</description></item></channel></rss>