{"version":1,"type":"story","url":"https://digestai.news/story/contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin","json":"https://digestai.news/story/contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin.json","markdown":"https://digestai.news/story/contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin.md","slug":"contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin","headline":"ContextAdapt evaluates LLMs on value alignment across medicine, law, finance, and national security","summary":"The paper introduces ContextAdapt, an evaluation framework to test whether large language models appropriately apply values like honesty, autonomy, and confidentiality across professional domains while remaining consistent when context does not change the relevant norm. Researchers evaluated 12 LLMs on recommended actions and justifications using scenarios based on primary-source professional and regulatory documents. In the main experiment, models achieved 95.6% mean appropriateness, but correct domain-specific justification varied from 25.6% to 76.9% across models. A separate factorial experiment found that naming the domain and changing the model's role had limited effect, while varying stakes revealed severe but localized failures, with models altering responses even when professional obligations were unchanged. Perceived severity acted as a cue for disclosure in honesty and confidentiality scenarios.","keyPoints":["ContextAdapt evaluates LLMs on honesty, autonomy, and confidentiality across medicine, law, finance, and national security","Models achieved 95.6% mean appropriateness but correct justification ranged from 25.6% to 76.9%","Varying stakes caused localized failures where models changed responses despite unchanged obligations"],"whyItMatters":"Shows that value alignment in LLMs requires contextual sensitivity beyond rule-following, highlighting risks in high-stakes professional use where models may misapply ethical principles.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":[],"people":[]},"firstPublishedAt":"2026-10-01T04:00:00Z","updatedAt":"2026-10-01T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.CL","title":"ContextAdapt: Evaluating Contextual Adaptation and Value Alignment in LLMs","url":"https://arxiv.org/abs/2609.38260","publishedAt":"2026-10-01T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"ContextAdapt evaluates LLMs on value alignment across medicine, law, finance, and national security\", 1 October 2026, https://digestai.news/story/contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin","publisher":"Digest AI","title":"ContextAdapt evaluates LLMs on value alignment across medicine, law, finance, and national security","datePublished":"2026-10-01T04:00:00Z","url":"https://digestai.news/story/contextadapt-evaluates-llms-on-value-alignment-across-medicine-law-fin"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}