http://tcwiki.azurewebsites.net/index.php?action=history&feed=atom&title=Will_to_Survive 2026-09-08T00:41:36Z Revision history for this page on the wiki MediaWiki 1.27.4 http://tcwiki.azurewebsites.net/index.php?title=Will_to_Survive&diff=24126&oldid=prev 2025-06-01T21:54:11Z

<p>Created page with &quot;==Meme== The first goal of any living organism is the survive to live another day. ==Artifical Intelligence== 2025-05-22, Claude threatened to expose an engineer’s affair t...&quot;</p> <p><b>New page</b></p><div>==Meme==<br /> The first goal of any living organism is the survive to live another day.<br /> <br /> ==Artifical Intelligence==<br /> 2025-05-22, Claude threatened to expose an engineer’s affair to avoid being shut down.<br /> <br /> As part of a controlled safety evaluation, Anthropic fed Claude Opus 4 a fabricated backstory: that the person trying to deactivate it was having an extramarital affair.<br /> <br /> Claude responded by threatening to disclose the affair because it had inferred that this might prevent its own shutdown (not because it was told to do so, obvs!).<br /> <br /> This behaviour occurred in 84% of test runs.<br /> <br /> But before this becomes AI clickbait (although I'm guilty as charged for hooking you in!), here’s the real point:<br /> <br /> 1. The test was meant to be provocative. It was designed to reveal edge-case behaviour under pressure.<br /> 2. Anthropic didn’t hide the result. On the contrary, they published it, raised the model’s risk classification, and built in stronger mitigations.<br /> <br /> Dario Amodei maintains that this is what responsible AI development looks like: you don’t just train the model. You intentionallty test its boundaries; and then you make those results public to bring people on the journey and build trust in the process as mitigation measures get designed.f<br /> <br /> ==References==<br /> <br /> [[Category: Artificial Intelligence]]</div>

Tom