<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai-Architecture on Terminal Wars</title><link>https://terminalwars.com/tags/ai-architecture/</link><description>Recent content in Ai-Architecture on Terminal Wars</description><generator>Hugo -- gohugo.io</generator><language>en</language><managingEditor>ender@terminalwars.com (Ender)</managingEditor><webMaster>ender@terminalwars.com (Ender)</webMaster><copyright>© 2013-2026 Terminal Wars</copyright><lastBuildDate>Fri, 26 Jun 2026 09:30:00 -0400</lastBuildDate><atom:link href="https://terminalwars.com/tags/ai-architecture/index.xml" rel="self" type="application/rss+xml"/><item><title>Securing Model Routing: When the Cheapest Path Picks the Guardrail</title><link>https://terminalwars.com/posts/model-routing/</link><pubDate>Fri, 26 Jun 2026 09:30:00 -0400</pubDate><author>ender@terminalwars.com (Ender)</author><guid>https://terminalwars.com/posts/model-routing/</guid><description>How LLM routers create cost efficiency and a novel attack surface when safety controls are tied to routing output rather than enforced as invariants across all routes.</description><media:content xmlns:media="http://search.yahoo.com/mrss/" url="https://terminalwars.com/posts/model-routing/featured.webp"/></item></channel></rss>