# faithfulness

Chain-of-Thought Faithfulness Attacks

Exploiting unfaithful chain-of-thought reasoning where models' stated reasoning does not match their actual decision process.