Safety Beyond the Interface: Detecting Harm via Latent States in Large Language Models.

Alizishaan Khatri, Chiquita Prabhu, Omkar Neogi

Browse the full DSN paper archive.