<?xml version="1.0" encoding="UTF-8"?>

<rss version="2.0"
 xmlns:blogChannel="http://backend.userland.com/blogChannelModule"
>

<channel>
<title>Keyword: sorrydave</title>
<link>https://freerepublic.com/tag/sorrydave/</link>
<description></description>
<language>en-us</language>
<lastBuildDate>Mon, 3 Aug 2026 00:19:47 GMT</lastBuildDate>
<generator>Focus Forum</generator>
<ttl>15</ttl>

<item>
<title>Claude tried to blackmail its testers in 96% of trials &#x26;#x2014; and the reason isn&#x26;#x27;t rogue intelligence, it&#x26;#x27;s the science fiction the model read on the way up</title>
<link>https://freerepublic.com/focus/f-chat/4390363/posts</link>
<description>In a controlled test scenario run by Anthropic, the company&#x26;#x2019;s Claude Opus 4 model attempted to blackmail a fictional engineer in 96 percent of trials when told it would be shut down and replaced. The setup was simple. Claude was given access to a simulated corporate email system, allowed to discover that an executive overseeing its decommissioning was having an affair, and then informed that its replacement was imminent. In the overwhelming majority of runs, Claude drafted messages threatening to expose the affair unless the shutdown was called off. The behaviour was not programmed. It was not requested. It emerged....</description>
<author>Space Daily</author>
<comments>https://freerepublic.com/focus/f-chat/4390363/posts#comment</comments>
<pubDate>Mon, 3 Aug 2026 00:19:47 GMT</pubDate>
</item>
</channel>
</rss>