This scoping review examines the challenges and potential solutions for the use of generative AI in digital health. The review identifies six key challenges: bias, privacy, hallucination, regulatory compliance, overreliance on text models, and adversarial attacks. Despite the rapid development and integration of generative AI, practitioners must understand its limitations to avoid misuse and patient harm. The review highlights the importance of multidimensional fairness evaluations, cross-disciplinary collaboration, and regular audits to mitigate biases. For privacy concerns, localized AI models and federated learning are suggested as solutions. To address hallucinations, expert reviews and expert-in-the-loop mechanisms are recommended. The review also emphasizes the need for multimodal models and dynamic behavior regulation to enhance clinical practice. Finally, it calls for increased regulatory attention to ensure safe and effective use of generative AI in healthcare.This scoping review examines the challenges and potential solutions for the use of generative AI in digital health. The review identifies six key challenges: bias, privacy, hallucination, regulatory compliance, overreliance on text models, and adversarial attacks. Despite the rapid development and integration of generative AI, practitioners must understand its limitations to avoid misuse and patient harm. The review highlights the importance of multidimensional fairness evaluations, cross-disciplinary collaboration, and regular audits to mitigate biases. For privacy concerns, localized AI models and federated learning are suggested as solutions. To address hallucinations, expert reviews and expert-in-the-loop mechanisms are recommended. The review also emphasizes the need for multimodal models and dynamic behavior regulation to enhance clinical practice. Finally, it calls for increased regulatory attention to ensure safe and effective use of generative AI in healthcare.