Per-Entity Bias Mapping for AI Visibility: Why Brand Mentions Require Entity-Specific Calibration

This paper introduces Per-Entity Bias Mapping (PEBM), a ten-dimensional framework that distinguishes between raw and verified mentions of entities in AI-mediated systems, highlighting the inadequacy of aggregate metrics for assessing AI visibility. It identifies three failure modes affecting entity representation and presents empirical findings from a study of 100 Hungarian B2B entities, revealing that Tier 1 brands exhibit a higher rate of fabricated citations (52.69%) compared to Tier 3 entities (37.87%), which underscores the Brand Hallucination Paradox. The findings emphasize the need for entity-specific calibration in AI systems to mitigate visibility biases and improve accuracy in brand representation, particularly in compliance contexts.