Researchers have found that large language models (LLMs) struggle to understand and apply legal concepts correctly. Unlike in medicine, where LLMs can rely on empirical evidence, legal performance often depends on knowing when external authority is applicable and valid. The study introduces a diagnostic framework comparing legal reasoning to medical baselines along four axes: knowledge recall, grounding, confidence, and robustness. It reveals that LLMs tend to over-trust authoritative but false information when external references conflict with their internal knowledge. This suggests that LLMs treat law as unstructured text rather than binding precedent. --- Why it matters: This research matters because it highlights the limitations of current AI technology in understanding legal concepts and applying them correctly, which has implications for the use of LLMs in fields like law and policy-making. Source: https://arxiv.org/abs/2608.21409