AI-TRUST: A Transparency Protocol for Generative AI-Assisted Research

Abstract

Generative artificial intelligence (GenAI) is transforming research, yet systematic evidence on disclosure remains limited. To address this gap, we systematically reviewed 257 articles published between 2023 and 2026 across 20 journals spanning general service management, hospitality, tourism and travel, retailing and consumer services, marketing, and business. GenAI disclosure quality was assessed across six dimensions: location of statement, tool specificity, purpose of use, prompt disclosure, human verification, and limitation acknowledgment. Findings revealed substantial weaknesses, particularly in the disclosure of prompts, verification procedures, and GenAI-specific limitations. Grounded in signaling theory, we propose AI-TRUST, a transparency protocol consisting of five core dimensions: traceability/auditability (T), reproducibility (R), understanding and verification (U), specification (S), and threat and limitation (T). AI-TRUST integrates an empirically derived six-variable transparency score with proportional reporting guidance to evaluate disclosure quality and strengthen procedural accountability.