PolyU researchers unveiled TRUECAM, a framework designed to show when pathology AI predictions are reliable versus uncertain, and when a human pathologist should take over. The system is built around the idea that clinical-grade AI must communicate uncertainty rather than only output classifications. The approach reflects a growing regulatory and clinical emphasis on trustworthy AI, where model performance depends not just on accuracy in aggregate but also on identifying failure modes at the patient level. TRUECAM is positioned as an operational tool for slide-based decision support, integrating uncertainty detection into the workflow. For health systems and pathology labs, the development adds a practical mechanism for human oversight in AI-enabled diagnostics, potentially supporting safer deployment and smoother integration into decision-making.