Ezra Newman(@EzraJNewman):Claude summarizes/describes behavior as significantly less misaligned when the actor is Claude vs another model In one experiment, Sonnet 5 describes the exact same data as ~1.2 std deviations less concerning when it describes misbehavior committed by Sonnet 5 vs GPT-5.6

Ezra Newman

@EzraJNewman

Member of Technical Staff at @ApolloResearch. Philosopher. Effective altruist. Likes != endorsement; epistemic status: twitter

参加 September 2020

434 フォロー中 1.4K ファン

Ezra Newman@EzraJNewman

2026.08.10 15:51

Claude summarizes/describes behavior as significantly less misaligned when the actor is Claude vs another model In one experiment, Sonnet 5 describes the exact same data as ~1.2 std deviations less concerning when it describes misbehavior committed by Sonnet 5 vs GPT-5.6