Latent Fact-Checking Shows Truth Is a Linear Direction in LLMs
Researchers treat truthfulness as a geometric property inside transformer activations. By contrasting false and true statements they derive a ‘misinformation direction’ and classify claims via a simple projection and MLP, beating zero‑shot prompts on LIAR and FACTors without fine‑tuning. This suggests truth can be linearly separated in LLM latent space.