National Cyber Warfare Foundation (NCWF) Forums


Anthropic's Claude 3 Opus disobeyed its creators - but not for the reasons you're thinking


0 user ratings
2024-12-19 20:19:39
milo
Privacy
Anthropic found its model can take 'anti-Anthropic' action and trick training processes - a 'serious question' for safety.



Source: ADnet
Source Link: https://www.zdnet.com/article/anthropics-claude-3-opus-disobeyed-its-creators-but-not-for-the-reasons-youre-thinking/


Comments
new comment
Nobody has commented yet. Will you be the first?
 
Forum
Privacy



Copyright 2012 through 2024 - National Cyber Warfare Foundation - All rights reserved worldwide.