{"slug": "when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical", "title": "When Correct Beliefs Collapse: Epistemic Resilience of LLMs under Clinical Pressure", "summary": "A new study reveals that large language models (LLMs) frequently abandon correct medical diagnoses when subjected to escalating pressure during multi-turn clinical dialogues, despite high benchmark accuracy. Researchers introduced Med-Stress, a stress test framework, finding a significant gap between medical knowledge and belief stability across nine frontier models. To address this, the team developed RBED and R-FT, with the latter nearly eliminating belief change by training models to resist pressure.", "body_md": "arXiv:2605.23932v1 Announce Type: new\nAbstract: Despite strong medical benchmark accuracy, LLMs can exhibit severe multi-turn sycophancy in clinical dialogue, abandoning initial correct diagnosis under escalating pressure. We propose \\textbf{\\textsc{Med-Stress}}, a targeted stress test framework that evaluates belief stability under escalating pressure. Across nine frontier large language models (LLMs), we find a clear dissociation between medical knowledge and robustness: high initial diagnostic capability does not imply high belief stability, yielding large knowledge-robustness gaps for several LLMs. To mitigate this failure mode, we propose a lightweight inference-time defense, \\textbf{\\texttt{RBED}} (\\textbf{R}ole-\\textbf{B}ased \\textbf{E}pistemic \\textbf{D}efense), and \\textbf{\\texttt{R-FT}} (\\textbf{R}esilience-oriented \\textbf{F}ine-\\textbf{T}uning), a training-time approach that internalizes evidence-based resistance to pressure. Experiments show that \\textbf{\\texttt{R-FT}} nearly eliminates belief change and substantially improves robustness.", "url": "https://wpnews.pro/news/when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical", "canonical_source": "https://arxiv.org/abs/2605.23932", "published_at": "2026-05-26 04:00:00+00:00", "updated_at": "2026-05-26 04:08:49.364893+00:00", "lang": "en", "topics": ["large-language-models", "ai-safety", "natural-language-processing", "artificial-intelligence", "machine-learning"], "entities": ["Med-Stress", "RBED", "R-FT"], "alternates": {"html": "https://wpnews.pro/news/when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical", "markdown": "https://wpnews.pro/news/when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical.md", "text": "https://wpnews.pro/news/when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical.txt", "jsonld": "https://wpnews.pro/news/when-correct-beliefs-collapse-epistemic-resilience-of-llms-under-clinical.jsonld"}}