1 paper toca esta idea.
Papers
AI models can't tell when a hacker put words in their mouth
Conceptos relacionados
Ideas que aparecen junto a LLM Introspection.