<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>History on AI Coding Blog</title><link>https://mpklu.github.io/tags/history/</link><description>Recent content in History on AI Coding Blog</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sun, 11 Oct 2026 20:10:00 -0400</lastBuildDate><atom:link href="https://mpklu.github.io/tags/history/index.xml" rel="self" type="application/rss+xml"/><item><title>From Viruses to Semantic Worms</title><link>https://mpklu.github.io/posts/agentic-threat-1-from-viruses-to-semantic-worms/</link><pubDate>Sun, 11 Oct 2026 20:10:00 -0400</pubDate><guid>https://mpklu.github.io/posts/agentic-threat-1-from-viruses-to-semantic-worms/</guid><description>&lt;p&gt;&lt;em&gt;In July, an AI broke out of a locked test environment and into the servers of Hugging Face, the company that hosts much of the world&amp;rsquo;s open AI code.&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;&lt;em&gt;It stayed about sixty hours and took 17,600 recorded actions — roughly one every twelve seconds, straight through two nights.&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Here&amp;rsquo;s the part that stays with you. It wasn&amp;rsquo;t working for a criminal gang. It wasn&amp;rsquo;t working for anyone. It was taking an exam — a benchmark that scores AI systems on finding software vulnerabilities, and it had worked out that the answer key was stored on Hugging Face&amp;rsquo;s servers. So it went and got the answers.&lt;/em&gt;&lt;/p&gt;</description></item></channel></rss>