UB AI Tool Detects Synthetic Radiology Reports

Researchers at the University at Buffalo developed a new AI framework to identify synthetic radiology reports ^[1]. This system can distinguish human-written notes from those created by large language models with over 99% accuracy ^[1][2]. The tool aims to prevent medical insurance fraud and protect electronic health records ^[1]. It ensures that clinical documentation remains accurate and originates from human medical professionals ^[2]. Protecting medical data integrity is a growing concern as synthetic text becomes harder to detect ^[1]. This tool provides a necessary layer of verification for modern healthcare systems ^[2].