<![CDATA[Medication administration errors can occur during preparation, verification, administration, and monitoring, threatening patient safety. Artificial intelligence (AI) may help nurses identify risks and medication mismatches, although technical performance does not always demonstrate clinical benefit. Objective: To review the use of AI to support medication administration safety and identify requirements for validation, human oversight, and integration into hospital nursing services. Methods: This systematic review with narrative synthesis followed PRISMA 2020. PubMed/PMC, Scopus, CINAHL, IEEE Xplore, and Google Scholar were searched for publications from 2020–2026. Of 1,425 records, 410 duplicates were removed, 1,015 records were screened, 53 full-text articles were assessed, and 13 articles were included, comprising five core empirical reports and eight supporting sources. Results: Evidence included machine-learning models for predicting error risk in neonatal care, deep-learning cameras for detecting vial swaps, and an AI–mixed-reality prototype for medication checking. One model reported an F1 score of 83.28%, while the camera system achieved 99.6% sensitivity and 98.8% specificity. Conclusion: AI may strengthen medication-safety checks, but it has not been proven to reduce errors or patient harm. Implementation requires local validation, AI literacy, human oversight, data protection, false-alert monitoring, and evaluation of workflow, nursing workload, and patient-safety outcomes through regular hospital monitoring.]]>
