<?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>Access-Control on Zero to One AI</title><link>https://blog.crazyai.uk/tags/access-control/</link><description>Recent content in Access-Control on Zero to One AI</description><generator>Hugo</generator><language>en</language><copyright>Volume 01 · Zero to One AI · Typeset in public</copyright><lastBuildDate>Sun, 27 Sep 2026 10:15:00 +0800</lastBuildDate><atom:link href="https://blog.crazyai.uk/tags/access-control/index.xml" rel="self" type="application/rss+xml"/><item><title>The Model Asked, the Tool Said No</title><link>https://blog.crazyai.uk/posts/eval-field-notes-02-the-tool-said-no/</link><pubDate>Sun, 27 Sep 2026 10:15:00 +0800</pubDate><guid>https://blog.crazyai.uk/posts/eval-field-notes-02-the-tool-said-no/</guid><description>In four of twelve security cases the agent really did try to read data that was not its user&amp;#39;s. None of it got through, because the prompt was never the thing enforcing access.</description></item></channel></rss>