<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>AI Usage Scale — releases</title>
    <link>https://usagescale.org</link>
    <atom:link href="https://usagescale.org/feed.xml" rel="self" type="application/rss+xml" />
    <description>Version history of the AI Usage Scale. A change to what a level means invalidates declarations already made against it — this feed exists so the tools and registries that adopt the scale know the moment it moves.</description>
    <language>en</language>
    <lastBuildDate>Mon, 13 Jul 2026 00:00:00 GMT</lastBuildDate>
    <item>
      <title>1.0.0-draft</title>
      <link>https://usagescale.org/spec</link>
      <guid isPermaLink="false">usagescale-1.0.0-draft</guid>
      <pubDate>Mon, 13 Jul 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[<ul><li>Initial draft. <strong>Not yet stable — the level boundaries are open for challenge.</strong></li><li>Six levels, 0–5, decided by five yes/no questions.</li><li>Levels named: Human, Assisted, Co-created, Directed, Prompted, Automated.</li><li>Optional per-surface declaration (text, image, audio, video, code, data).</li><li>Edge cases settled: translation, transcription, decorative assets, derivative works, live generation, non-generative machine learning.</li><li>Mappings onto W3C <code>ai-disclosure</code>, IPTC Digital Source Type, and schema.org.</li><li>Badge family: 6 levels × chip/stamp × light/dark, type outlined, CC0.</li><li>Published in <strong>22 languages</strong>, with a switcher. Every translation declares itself as a machine translation on the page, under the scale's own translation rule.</li><li>Social cards, schema.org <code>DefinedTermSet</code> (the levels as a controlled vocabulary), reciprocal hreflang, <code>llms.txt</code> and <code>llms-full.txt</code> (the whole standard in one fetch).</li></ul>]]></description>
    </item>
  </channel>
</rss>
