Suicide and self-harm represent critical public health challenges requiring innovative early detection and intervention strategies. Recent advances in artificial intelligence, particularly machine learning and natural language processing, have demonstrated promising potential for identifying individuals at elevated risk for suicidal behavior and self-injury. This narrative review synthesizes evidence regarding AI-assisted risk detection systems in mental health contexts, examining their clinical applications, empirical evidence, limitations, and implications for psychiatric and mental health nursing practice. Key findings indicate that machine learning models trained on electronic health record data, combined with natural language processing analysis of clinical narratives, can identify suicide risk with sensitivity ranging from 0.56 to 0.88 and specificity from 0.57 to 0.99, depending on model architecture and population characteristics. Hierarchical machine learning approaches integrating clinical history with repeated nursing assessments demonstrate superior performance for detecting self-harm risk in hospitalized psychiatric patients. However, successful clinical implementation requires addressing critical challenges including algorithmic bias, data privacy concerns, clinical validation across diverse populations, and preservation of the therapeutic nurse-patient relationship. Mental health nurses are positioned as essential stakeholders in AI implementation, requiring education regarding AI capabilities and limitations, meaningful participation in system design and governance, and clear accountability frameworks. This review concludes that responsible AI integration in suicide and self-harm risk detection must prioritize ethical implementation, maintain human clinical judgment as primary, and ensure that technology augments rather than replaces nursing assessment and therapeutic intervention.