News

How dangerous is encoded reasoning?

  • Artyom Karpov--Lesswrong.com
  • published date: 2025-06-30 11:54:12 UTC

Published on June 30, 2025 11:54 AM GMTEncoded reasoning occurs when a language model (LM) agent hides its true reasoning inside its chain-of-thought (CoT). It is one of the three types of unfaithful reasoning[1] and the most dangerous one because it undermin…

Encoded reasoning occurs when a language model (LM) agent hides its true reasoning inside its chain-of-thought (CoT). It is one of the three types of unfaithful reasoning[1] and the most dangerous on… [+23850 chars]