<?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>Compliance on westerweel.work</title><link>https://westerweel.work/en/tags/compliance/</link><description>Recent content in Compliance on westerweel.work</description><generator>Hugo</generator><language>en</language><lastBuildDate>Thu, 18 Jun 2026 21:43:38 +0200</lastBuildDate><atom:link href="https://westerweel.work/en/tags/compliance/index.xml" rel="self" type="application/rss+xml"/><item><title>Ricin breaks your model. Not your data.</title><link>https://westerweel.work/en/posts/2026-06-17-output-safety-is-not-data-confidentiality/</link><pubDate>Wed, 17 Jun 2026 00:00:00 +0000</pubDate><guid>https://westerweel.work/en/posts/2026-06-17-output-safety-is-not-data-confidentiality/</guid><description>Why the most popular AI safety test secures the front door while leaving the back door wide open. Output safety and data confidentiality are technically two completely different mechanisms — and conflating them is exactly what&amp;rsquo;s behind the confusion on the work floor. Here&amp;rsquo;s how models actually handle your data, and what to arrange instead of the theatre.</description></item></channel></rss>