<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>oopscompiled</title><description>Personal engineering blog about ML, AI systems and security.</description><link>https://oopscompiled.github.io/</link><item><title>Catch the Jailbreak If You Can</title><link>https://oopscompiled.github.io/blog/catch-the-jailbreak/</link><guid isPermaLink="true">https://oopscompiled.github.io/blog/catch-the-jailbreak/</guid><description>Jailbreak is not a break-in. It is Frank Abagnale for neural networks: why LLMs trust the costume more than the badge.</description><pubDate>Wed, 22 Jul 2026 00:00:00 GMT</pubDate></item><item><title>The Illusion of Control: Why Prompts Don&apos;t Protect AI Agents</title><link>https://oopscompiled.github.io/blog/illusion-of-control/</link><guid isPermaLink="true">https://oopscompiled.github.io/blog/illusion-of-control/</guid><description>From Permission Boundary Bypass to Dyck languages: why agent security has to live in the runtime, not in the system prompt.</description><pubDate>Tue, 23 Jun 2026 00:00:00 GMT</pubDate></item></channel></rss>