Toward Autonomous PFAS Forensics: AI‑Enhanced HRMS Workflows for Sustainable Chemical and Water System Management

Abstract PFAS contamination has become deeply embedded in the food−water system, creating a growing challenge for chemical safety, environmental protection, and sustainable resource management. Yet most current monitoring tools still rely on targeted methods that detect only a small fraction of the thousands of PFAS in circulation. As a result, major gaps remain in how we assess exposure, track environmental behavior, and design effective mitigation strategies. This perspective highlights how combining high resolution mass spectrometry (HRMS) with Artificial Intelligence (AI) can fundamentally transform PFAS monitoring. HRMS offers a comprehensive, non-targeted view of the chemical landscape in wastewater and food matrices, while machine learning and deep learning models facilitate the interpretation of these complex datasets. Together, they enable the discovery of previously unrecognized PFAS, improve predictions of their fate and transport, and support more accurate source attribution across food supply chains. We demonstrate how AI can reduce the analytical E-factor by minimizing solvent and energy use but highlight the paradoxical increase in embodied carbon associated with computational model training, a trade-off that requires systematic lifecycle evaluation. Framing the food-water nexus as a critical testing ground, we outline how these tools must evolve from forensic detectors into instruments of pollution prevention and circular resource management. Finally, we propose a focused research agenda prioritizing explainable AI for regulatory trust, real-time sensing for preventive control, and generative AI for designing safer alternatives. This critical assessment aims to steer the development of AI-HRMS from a powerful analytical technique into a cornerstone of sustainable chemical management.