Background: Clinical documentation is a major contributor to administrative workload in hospital settings. Increasing documentation requirements, combined with limited time and staffing resources, place substantial pressure on healthcare professionals. AI-enabled digital health technologies (DHTs) are increasingly proposed to support or partially automate documentation tasks, aiming to reduce documentation burden and improve workflow efficiency. From a regulatory perspective, most AI-based documentation tools are currently classified as low-risk medical devices or non-medical software. Nevertheless, their use raises important questions regarding accuracy, completeness, data protection, bias, accountability, and governance. This report presents a scoping review that maps key AI-enabled documentation support functions, and describes the evidence base for their performance and impact in hospitals.

Methods: A systematic literature search was conducted across four bibliographic databases. In total, 755 records were screened. AI-supported documentation functions were grouped into use cases, and the evidence was synthesised narratively with a focus on technical performance, reported benefits, limitations, and implementation considerations.

Results: Seven reviews (three systematic and four scoping reviews), comprising approximately 200 primary studies, were included. Six use cases were identified: AI scribes, text structuring, AI-generated documentation, patient-friendly summaries, error detection, and automated billing code assignment. The evidence base was heterogeneous across use cases. AI scribes were most frequently studied, showing variable accuracy and completeness, with omissions commonly reported. Clinician satisfaction was generally high, but evidence on time savings and productivity gains was inconsistent. AI-generated documentation and text structuring showed potential benefits but required human oversight due to omissions and hallucinations. Automated billing code assignment and error detection demonstrated high technical accuracy, while patient-friendly summaries improved patient comprehension. Across use cases, data protection concerns, legal uncertainties, and limited evidence on organisational outcomes were repeatedly identified.

Discussion: Evidence on AI-enabled documentation support remains uneven and dominated by technical evaluations. Organisational impacts, implementation requirements, and long-term performance are insufficiently studied. Given the potential influence of these tools on clinical records, implementation requires human oversight, local validation, and governance frameworks, particularly in the Austrian context with ELGA and increasing interoperability requirements.

Conclusion: AI-enabled documentation tools have the potential to reduce administrative burden; however, current evidence is limited, inconsistent, and highly context-dependent. A proportionate, risk-based approach with structured validation, sustained human oversight, and governance mechanisms is required before large-scale deployment.