Muhan Luo's Shortform
Aug 122
Muhan Luo
Message
LinkedIn: https://www.linkedin.com/in/muhan-luo-620259171/
Substack: https://muhanluo.substack.com/
15
2
3
TL;DR: Anthropic restricted access to Claude Mythos Preview, citing a major leap in vulnerability discovery and exploitation capability. I review the 3 most common arguments from skeptics: (1) AISLE Security’s paper showing cheaper models can identify the same bugs as Mythos, (2) benchmark comparisons showing GPT-5.5 performs comparably, and (3)...