Anirandis comments on There’s probably a tradeoff between AI capability and safety, and we should act like it