MOSAIC Benchmark Evaluates LLM Moral, Social, and Individual Dimensions
August 14, 2026
MOSAIC introduces a large-scale benchmark using nine validated questionnaires and four platform-based games to assess LLM behavior across moral philosophy, psychology, and social theory. It moves beyond standard Moral Foundation Theory to evaluate how personality traits and social values influence ethical reasoning.
HOW THIS AFFECTS YOU
●
researcherYou can use this to move beyond MFT-only evaluations in ethical reasoning studies.
●
policyThis provides more nuanced metrics for assessing model reliability in high-stakes social roles.