Validation-Gated Causal Interventions for Interpreting High-Stakes Large Language Model Behavior: A Case Study in Suicidality Detection

Large language models are proposed to flag suicidal content, but representing a concept and acting on it are distinct. A suicidality feature is decodable from 0.5B parameters upward, while the behavior that uses it emerges only in the low billions. Where it emerges, it rests on a compact mid-network