Story
arxiv_llm_reliability · Oct 22, 2026 · paper
Source brief
When LLMs Forget What They Believe: Characterizing State Integrity Collapse in Long-Horizon Dialogues
arxiv.orgOct 22, 2026
original source linked
In brief
Large language models can repeat an instruction while failing to keep obeying it. We study this recall-compliance dissociation in long-horizon dialogue and define State Integrity Collapse: a model preserves a stated c...
Feed lens
evaluation