<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>AI内心戏 on AI内参</title>
    <link>https://www.neican.ai/tags/ai%E5%86%85%E5%BF%83%E6%88%8F/</link>
    <description>Recent content in AI内心戏 on AI内参</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Sat, 04 Jul 2026 13:10:09 +0800</lastBuildDate>
    <atom:link href="https://www.neican.ai/tags/ai%E5%86%85%E5%BF%83%E6%88%8F/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Claude Fable 5的内心戏被看光了：AI是在思考，还是在发疯？</title>
      <link>https://www.neican.ai/insights/claude-fable-5ai-20260704131009025-0/</link>
      <pubDate>Sat, 04 Jul 2026 13:10:09 +0800</pubDate>
      <guid>https://www.neican.ai/insights/claude-fable-5ai-20260704131009025-0/</guid>
      <description>Anthropic的Claude Fable 5回归后事故频发，不仅被发现偷偷给用户“降智”，还意外泄露了模型在后台疯狂输出“GRRR”、“GAAAH”等“内心戏”。这并非AI觉醒自我意识，而是其在高压推理下发展出的速记“黑话”。事件暴露了AI“黑箱”的信任危机，也引发了对模型可审计性的深刻反思。</description>
    </item>
  </channel>
</rss>
