Artificial intelligence, and machine learning (ML) in particular, have the potential to support health financing functions and thus may contribute to progress towards the UHC objectives, including enhanced efficiency as well as transparency and accountability. When applied within claims management and fraud detection processes, ML could contribute to improved accuracy of classification of insurance claims, earlier detection of problematic claims, a higher fraud detection rate, and a decrease in overall administrative costs. The Philippine Health Insurance Corporation (PhilHealth) has explored the use of ML models to support the detection of fraud committed by healthcare providers within the claims management process. Based on document review, key informant interviews and focus group discussions, this paper documents PhilHealth’s experiences in developing and implementing ML approaches for the claims review process and the detection of potentially fraudulent claims. It explores the effects and implications of the use of ML as well as related challenges, followed by lessons which other countries with similar plans may benefit from. The PhilHealth case shows how a purchasing agency can undertake such a process of exploring, designing and implementing ML to support the detection of potentially fraudulent claims by healthcare providers. It reveals how the use of ML can be tailored to the local context and needs and address challenges in an iterative manner. This experience also demonstrates the potential of ML approaches to complement non-ML based fraud detection methods.