Artificial Intelligence in Nursing Governance and Regulation: An Umbrella Review of Ethical and Policy Dimensions

Abstract

Background

As AI becomes increasingly embedded in healthcare systems, nursing governance faces new challenges involving ethical accountability, professional autonomy, data stewardship, and institutional oversight. Existing reviews highlight fragmented understanding of how these changes impact the nursing profession.

Aim

This umbrella review aimed to synthesize ethical and policy dimensions related to the integration of artificial intelligence (AI) within nursing governance and regulatory frameworks.

Methods

Following JBI guidance and PRISMA 2020, five databases were searched for reviews published from January 2010 to December 2025. Reviews were appraised and synthesized by purpose, quality, nursing specificity, and primary study overlap, which was quantified using a citation matrix and Corrected Covered Area (CCA).

Findings

Thirty-one reviews included 23 evidence syntheses and eight evidence maps. Privacy or data stewardship appeared in 29 reviews, transparency or explainability in 27, bias or fairness in 25, accountability or liability in 25, consent or autonomy in 18, leadership, education, or oversight in 14, and safety or human oversight in nine. Among 23 reviews with enumerable, extractable study lists, the CCA was 0.54%, indicating slight overlap. Nursing-specific concerns involved professional judgment, representation, oversight, regulatory variation, and ethical preparedness.

Conclusion

AI creates linked governance concerns involving data, fairness, transparency, autonomy, and accountability. Auditable responsibilities are needed across clinical, institutional, and regulatory levels while preserving nursing judgment and patient advocacy. Evidence for specific regulatory models remains limited.

Introduction

Artificial intelligence (AI) is transforming clinical, administrative, and governance processes (Wubineh et al., 2024) and changing nursing as both a caregiving discipline and a regulated profession (Bodur et al., 2025). Despite its benefits for diagnostic accuracy, workflow, and prediction, its governance implications remain insufficiently examined (Abdelmohsen & Al-jabri, 2025). Nursing governance comprises policies, leadership structures, accountability mechanisms, ethical standards, and regulations that guide practice (Cao et al., 2025), including professional accountability and autonomy in digital care (Almagharbeh et al., 2025; George & Peirce, 2025). AI-supported decision-making and workforce systems reshape structures grounded in autonomy, advocacy, and stewardship (Parry et al., 2023; Rony et al., 2024), potentially constraining clinical judgment (Akter et al., 2025). Frameworks designed for human-led decisions should therefore address new boundaries of accountability, transparency, and responsibility (Mohammad Amini et al., 2023; Nesa et al., 2025; Secinaro et al., 2021).

As a policy actor, AI affects decisions, resources, professional responsibilities, and organizational governance (Mikkonen et al., 2025), requiring safeguards that preserve nursing values (Yang & Ricciardi, 2026). Privacy, algorithmic bias, professional liability, and informed consent are prominent concerns (Witkowski et al., 2024), particularly when limited nursing participation separates governance from frontline realities (Arcadi, 2025), contributes to policy fragmentation (Gesser Da Costa et al., 2025), and compounds regulatory variation (Park, 2025). Nurses consequently require ethical preparedness to evaluate and govern AI responsibly (Tun et al., 2025). This governance perspective is especially important because nurses encounter AI throughout the care pathway rather than during a single technical decision (Mohammed et al., 2025; O’Connor et al., 2023). They may supply data, interpret predictions, communicate recommendations, document actions, coordinate escalation, and monitor outcomes, with each stage creating distinct professional and institutional responsibilities (Mikkonen et al., 2025).

Governance should therefore address not only algorithmic accuracy but also whether nurses receive adequate information, authority, training, and protection when questioning AI-generated outputs (Ruksakulpiwat et al., 2024). It should also consider how procurement, vendor responsibilities, data sharing, and post-deployment monitoring affect bedside practice. Connecting these issues distinguishes nursing governance from broader healthcare AI ethics and helps determine whether existing professional standards remain suitable when authority is distributed among nurses, healthcare organizations, technology developers, and automated systems (Park, 2025; Van Der Gaag et al., 2023). Such governance must keep patient advocacy and relational care visible in technology-related policy decisions (Arcadi, 2025). This umbrella review synthesized the ethical and regulatory implications of AI for nursing governance, professional roles, policy, and emerging risks to inform coherent, equitable, and patient-centered governance.

Methodology

Review Design and Framework

This umbrella review synthesized systematic, qualitative systematic, mixed-methods, integrative, rapid, scoping, and meta-synthesis reviews of ethical and policy issues in AI-related nursing governance. Procedures followed JBI recommendations (Hilton, 2024) and PRISMA reporting (Page et al., 2021) (Figure 1); the checklist is in Supplementary file 1. The protocol was registered in PROSPERO (Registration No. CRD420261332265). The review was the unit of inclusion. Evidence syntheses consolidated findings, while evidence maps identified breadth and gaps. Scoping reviews followed their mapping purpose (Arksey & O’Malley, 2005; Peters et al., 2020) and were not treated as systematic reviews of effects.

Figure 1.

PRISMA flow diagram

Objectives and Review Questions

The objective was to explore how AI intersected with ethical principles and regulatory structures in nursing governance and to identify overarching themes, policy challenges, and ethical considerations. The review asked: What ethical and policy dimensions of AI applications in nursing governance and regulation were reported in systematic or scoping reviews? Sub-questions addressed privacy, accountability, bias, oversight, autonomy, and leadership.

Eligibility Criteria

Eligible English-language systematic reviews, scoping reviews, or meta-syntheses published from January 2010 to December 2025 examined AI or machine learning in nursing or broader healthcare governance, ethics, or policy and addressed ethical principles or regulatory matters. Primary studies, editorials, non-peer-reviewed opinions, and publications limited to clinical AI without ethical or governance content were excluded. Eligible labels included systematic, qualitative systematic, mixed-methods, integrative, rapid, scoping, and meta-synthesis reviews. Inclusion required reproducible methods and extractable nursing-relevant governance, ethics, accountability, policy, or regulatory findings. Education, attitude, specialty-care, or clinical reviews provided contextual evidence only when they explicitly addressed autonomy, consent, liability, oversight, data stewardship, fairness, transparency, safety, or nursing participation.

Information Sources and Search Strategy

PubMed, Scopus, CINAHL, Web of Science, and IEEE Xplore contributed records to the PRISMA flow. Search terms covered AI, machine learning, generative AI, large language models, algorithmic decisions, decision support, predictive analytics, robotics, digital health, nursing informatics, governance, ethics, policy, regulation, licensure, scope of practice, liability, oversight, accountability, and review designs (Table 1). Boolean, truncation, field, and proximity operators were adapted by database. Searches covered January 2010 to December 2025 and were executed between January 10 and February 20, 2026. Grey literature was ineligible and absent from PRISMA counts. Full syntax, fields, filters, limits, and available dates are reported in the supplementary search appendix.

Table 1.

Search Strategy

Database	Database-specific syntax	Search conducted	Coverage	Limits
PubMed	(“artificial intelligence” OR “machine learning” OR “deep learning” OR “generative AI” OR “large language model*” OR “algorithmic decision-making” OR “clinical decision support” OR “predictive analytics” OR robotic* OR “digital health”) AND (nurs* OR “nursing informatics”) AND (govern* OR ethic* OR polic* OR regulat* OR licens* OR “scope of practice” OR liabilit* OR oversight OR accountab* OR transparen* OR explainab* OR privacy OR bias OR fairness) AND (“systematic review” OR “scoping review” OR “integrative review” OR “rapid review” OR meta-synth*) [Title/Abstract; MeSH terms adapted where available]	January 10–February 20, 2026	Jan 2010–Dec 2025	English; review-level publications
CINAHL	(“artificial intelligence” OR “machine learning” OR “deep learning” OR “generative AI” OR “large language model*” OR “algorithmic decision-making” OR “clinical decision support” OR “predictive analytics” OR robotic* OR “digital health”) AND (nurs* OR “nursing informatics”) AND (govern* OR ethic* OR polic* OR regulat* OR licens* OR “scope of practice” OR liabilit* OR oversight OR accountab* OR transparen* OR explainab* OR privacy OR bias OR fairness) AND (“systematic review” OR “scoping review” OR “integrative review” OR “rapid review” OR meta-synth*) [TX/MH fields adapted]	January 10–February 20, 2026	Jan 2010–Dec 2025	English; review-level publications
Scopus	TITLE-ABS-KEY((“artificial intelligence” OR “machine learning” OR “deep learning” OR “generative AI” OR “large language model*” OR “algorithmic decision-making” OR “clinical decision support” OR “predictive analytics” OR robotic* OR “digital health”) AND (nurs* OR “nursing informatics”) AND (govern* OR ethic* OR polic* OR regulat* OR licens* OR “scope of practice” OR liabilit* OR oversight OR accountab* OR transparen* OR explainab* OR privacy OR bias OR fairness) AND (“systematic review” OR “scoping review” OR “integrative review” OR “rapid review” OR meta-synth*))	January 10–February 20, 2026	Jan 2010–Dec 2025	English; review-level publications
Web of Science	TS=((“artificial intelligence” OR “machine learning” OR “deep learning” OR “generative AI” OR “large language model*” OR “algorithmic decision-making” OR “clinical decision support” OR “predictive analytics” OR robotic* OR “digital health”) AND (nurs* OR “nursing informatics”) AND (govern* OR ethic* OR polic* OR regulat* OR licens* OR “scope of practice” OR liabilit* OR oversight OR accountab* OR transparen* OR explainab* OR privacy OR bias OR fairness) AND (“systematic review” OR “scoping review” OR “integrative review” OR “rapid review” OR meta-synth*))	January 10–February 20, 2026	Jan 2010–Dec 2025	English; review-level publications
IEEE Xplore	(“artificial intelligence” OR “machine learning” OR “deep learning” OR “generative AI” OR “large language model*” OR “algorithmic decision-making” OR “clinical decision support” OR “predictive analytics” OR robotic* OR “digital health”) AND (nurs* OR “nursing informatics”) AND (govern* OR ethic* OR polic* OR regulat* OR licens* OR “scope of practice” OR liabilit* OR oversight OR accountab* OR transparen* OR explainab* OR privacy OR bias OR fairness) AND (“systematic review” OR “scoping review” OR “integrative review” OR “rapid review” OR meta-synth*)	January 10–February 20, 2026	Jan 2010–Dec 2025	English; review-level publications

Study Selection Process

Records were imported into Zotero and deduplicated. Two reviewers independently screened titles, abstracts, and full texts; disagreements were resolved through discussion or a third reviewer. Full-text exclusions and reasons were recorded, and the selection process was summarized in the PRISMA diagram. Quality was not an eligibility criterion. Full-text exclusions reflect prespecified publication-type, population, AI, nursing, or governance criteria; the PRISMA diagram therefore uses eligibility-based rather than quality-based reasons.

Data Extraction

A piloted form captured citation, setting, review design, included-study characteristics, AI technologies, ethical and regulatory topics, theoretical or policy approaches, and findings. Two reviewers extracted and cross-checked data independently and contacted authors when clarification was required.

Quality Appraisal of Included Reviews

Two reviewers independently applied JBI criteria to systematic or research syntheses and JBI-informed criteria to scoping reviews (Supplementary file 2), resolving disagreements through discussion. “Yes” was scored 1, “partial” 0.5, and “no” or “unclear” 0. Of the 31 included reviews, 22 (71.0%) were rated high quality, six (19.3%) moderate quality, and three (9.7%) low quality. Reviews were not excluded by quality. Overall categories were interpreted with item-level judgments, and weaknesses in independent appraisal, extraction-error minimization, publication-bias assessment, or other critical domains prevented review from independently establishing strong convergence.

Data Synthesis and Analysis

Methodological and conceptual heterogeneity precluded meta-analysis, so narrative thematic synthesis was used. Systematic reviews informed qualified confirmation, while scoping reviews mapped emerging issues. Reviewers iteratively coded transparency, liability, algorithmic governance, oversight, and related concepts, reconciled themes by consensus, and mapped them into ethical frameworks and policy typologies. Reviews were classified as evidence synthesis, evidence mapping, or exploratory synthesis, and themes were counted once per review. Primary-study overlap was assessed using a review-by-study citation matrix, with DOI as the primary identifier and unambiguous first author/year matching when DOI was unavailable. CCA was calculated as (N − r)/(r × c − r), where N was the number of study occurrences, r the number of unique studies, and c the number of reviews analyzed; 0%–5% indicated slight overlap (Pieper et al., 2014). Complete, extractable study lists were available for 23 reviews; eight were not represented in this analysis.

Results

Summary of Included Studies

This umbrella review synthesized evidence from 31 reviews examining the application and governance of artificial intelligence (AI) across nursing practice, education, management, and broader healthcare systems (Supplementary file 3). The reviews reflected a global perspective, encompassing North America, Europe, Asia, the Middle East, Australia, and selected low- and middle-income countries. Hospital and acute care settings predominated, although community care, long-term care, nursing education, and healthcare administration were also represented. Methodologically, the evidence base included systematic, scoping, integrative, and rapid reviews guided by frameworks such as PRISMA, the Joanna Briggs Institute, and Arksey and O’Malley. Study populations included nurses, students, educators, policymakers, and patients, while AI applications ranged from clinical decision support systems and predictive analytics to machine learning, natural language processing, robotics, and generative AI. Although the search covered publications from 2010 to 2025, all eligible reviews were published between 2020 and 2025, reflecting the recent emergence of scholarship focused on AI governance, ethics, and regulation in nursing. Systematic reviews generally provided stronger evidence than scoping reviews, which primarily mapped emerging themes. Using the prespecified thresholds, 22 reviews were rated high quality, six moderate quality, and three low quality (Supplementary file 2). For interpretation, systematic and other evidence-synthesis reviews contributed to qualified convergence, whereas scoping and other evidence-mapping reviews identified emerging domains and gaps. Reviews directly focused on nursing governance or ethics were distinguished from broader clinical, educational, or healthcare reviews that supplied contextual governance findings. Lower-quality reviews were used to identify possible concerns but could not independently establish convergence.

Review-level support comprised privacy or data stewardship (n=29), transparency or explainability (n=27), bias or fairness (n=25), accountability or liability (n=25), consent or autonomy (n=18), leadership, education, or oversight (n=14), and safety or human oversight (n=9). Across the 23 reviews included in the overlap analysis, 694 study occurrences represented 620 unique primary studies, yielding 74 repeated occurrences and a CCA of 0.54% (slight overlap). Sixty-two studies appeared in at least two reviews, with a maximum recurrence of five reviews (Supplementary file 4).

Ethical Challenges in AI Integration Across Nursing Governance

A prominent theme across the included reviews was the ethical uncertainty surrounding the integration of artificial intelligence into nursing governance structures (Figure 2). While AI offers opportunities to enhance decision-making, reduce administrative burden, and improve patient outcomes (Burford et al., 2025), it simultaneously introduces challenges related to autonomy, beneficence, justice, and non-maleficence (Kiwanuka et al., 2025). Several reviews highlighted concerns about preserving nurse–patient relationships in increasingly algorithm-driven care environments (Mohammed et al., 2025; Pereira et al., 2025). The mechanization of workflows through decision support systems and automated triage may constrain clinical judgment and reduce professional autonomy (Labrague & Al Harrasi, 2025). Additionally, reliance on AI-generated recommendations raises the risk of overtreatment, undertriage, or neglect of patient preferences (Ramírez-Baraldes et al., 2025). Ethical frameworks traditionally guiding nursing practice may not sufficiently address these emerging complexities (Mikkonen et al., 2025). Although principles such as autonomy, beneficence, non-maleficence, and justice remain central, they require reinterpretation in contexts where decisions are influenced by opaque algorithmic processes (Gonzalez-Garcia et al., 2024). Limited transparency in AI systems further challenges informed consent and shared decision-making (O’Connor et al., 2024), underscoring the need to redefine ethical accountability within both professional and institutional domains (Cao et al., 2025).

Figure 2.

Review-level AI governance domains in nursing

Algorithmic Bias and Fairness in AI Policy Discourse

Another consistent finding across the included reviews was the presence of algorithmic bias as a significant threat to equity in nursing governance. Many AI systems are developed using datasets that lack diversity or reflect historical inequalities in healthcare delivery, thereby reproducing or amplifying discriminatory outcomes (El Arab et al., 2025a). For instance, predictive analytics used for patient risk stratification or resource allocation may embed biases related to race, gender, or socioeconomic status (Koo et al., 2024). From a regulatory perspective, few reviews reported comprehensive mechanisms to evaluate, monitor, or mitigate such biases at the institutional level (Buchanan et al., 2020; Pereira et al., 2025). This gap between technological advancement and governance preparedness is particularly concerning in multicultural care environments where equity is central (Koo et al., 2024). Algorithmic fairness remains underdeveloped within nursing regulatory frameworks, with limited evidence of standardized bias auditing or accountability processes (Ramírez-Baraldes et al., 2025). If left unaddressed, these biases risk exacerbating healthcare disparities and undermining core ethical commitments to justice and inclusivity (Burford et al., 2025).

Data Privacy, Consent, and Information Governance

Data governance emerged as a central concern across the included reviews, particularly regarding the collection, sharing, and use of data within AI-driven nursing systems. The increasing reliance on large-scale datasets has blurred the boundaries between personal and professional data, raising significant concerns for both patient and nurse privacy (Porcellato et al., 2025). Many AI applications depend on de-identified data; however, the effectiveness of de-identification remains contested due to the potential for re-identification through advanced analytical techniques (Karimian et al., 2022). These risks complicate traditional informed consent processes, which often fail to capture the dynamic and continuous nature of AI-enabled data use (Ventura-Silva et al., 2024). Furthermore, institutional policies frequently lag behind technological capabilities, leaving gaps in the governance of sensitive nursing data such as documentation patterns, workflow metrics, and behavioral data from wearable technologies (Ruksakulpiwat et al., 2024). Questions surrounding data ownership, storage, and secondary use remain insufficiently addressed, highlighting the urgent need for clearer regulatory frameworks and stronger data stewardship practices (Buchanan et al., 2020).

Professional Accountability in Algorithm-Supported Decision-Making

The integration of AI into clinical and administrative processes has raised critical concerns regarding professional accountability within nursing governance. Several reviews highlighted situations where AI-generated recommendations conflict with nurses’ clinical judgment, creating uncertainty about responsibility in decision-making (Pereira et al., 2025; Porcellato et al., 2025). This ambiguity is particularly evident when adverse outcomes occur, as it becomes unclear whether accountability lies with the nurse, the institution, or the AI system itself (O’Connor et al., 2024). Nurses may feel compelled to rely on algorithmic outputs, especially when institutional policies mandate their use, potentially limiting independent clinical reasoning (Mikkonen et al., 2025). At the same time, traditional regulatory frameworks and licensure standards have not yet evolved to accommodate AI-supported decision-making processes (O’Connor et al., 2023). This gap increases the risk of moral distress and legal vulnerability among nurses (Mohammed et al., 2025). Furthermore, limited training in interpreting AI outputs restricts nurses’ ability to critically evaluate these systems, highlighting the need for clearer accountability guidelines (Moghadam et al., 2024).

Transparency and Explainability of AI Systems

Transparency and explainability were consistently identified as essential requirements for ethical AI integration in nursing governance. Many reviews emphasized that AI systems, particularly those based on deep learning and complex models, often function as “black boxes,” limiting understanding of how outputs are generated (Al Khatib & Ndiaye, 2025; Karimian et al., 2022). This lack of interpretability reduces trust among nurses and constrains their ability to appropriately act on AI-driven recommendations (Von Gerich et al., 2022). It also challenges ethical obligations related to patient communication, informed consent, and shared decision-making (Mikkonen et al., 2025). Several studies highlighted that explainability should be viewed not only as a technical feature but also as a governance principle that supports accountability and oversight (Katebi et al., 2025; Ruksakulpiwat et al., 2024). However, current regulatory frameworks rarely mandate explainability as a requirement for AI adoption (El Arab et al., 2025b). Without transparency, the risk of misuse, misinterpretation, and resistance to AI technologies increases, particularly in settings lacking clinician involvement (O’Connor et al., 2023).

Institutional Oversight, Leadership, and Policy Dimensions in AI Governance

Institutional fragmentation and limited oversight were consistently identified as barriers to the ethical implementation of AI in nursing governance. Reviews indicated that AI initiatives are frequently introduced without coherent governance structures, resulting in inconsistent implementation and poor alignment with nursing values (Koo et al., 2024; Martinez-Ortigosa et al., 2023). Although national digital health strategies encourage innovation, local governance mechanisms often remain underdeveloped (Buchanan et al., 2020). AI adoption is commonly driven by efficiency and cost-reduction goals, with limited involvement of nurse leaders or ethical oversight bodies (Porcellato et al., 2025). Consequently, AI-related decisions may be implemented without adequate consideration of their implications for nursing practice and professional values (Pereira et al., 2025). Limited nursing representation within interdisciplinary ethics committees further restricts opportunities for ethical deliberation (O’Connor et al., 2024).

Nursing leadership was identified as a critical yet underdeveloped component of AI governance. Many nurse leaders lack sufficient training and institutional support to participate effectively in AI policymaking (Ruksakulpiwat et al., 2024), increasing the risk that nurses become passive users rather than active contributors to AI implementation (Mohammed et al., 2025). Promising strategies include AI literacy programs, integration of ethical competencies into nursing education, and stronger collaboration between nursing organizations and regulatory bodies (Alqaissi & Qtait, 2025). Accordingly, a multi-level AI governance framework is proposed (Figure 3), emphasizing accountability, transparency, fairness, data stewardship, and nursing leadership across clinical, institutional, and regulatory levels.

Figure 3.

Conceptual framework for artificial intelligence governance in nursing. (Note: This figure is an author-developed conceptual proposal informed by the completed thematic synthesis. It was not generated through a separate inductive theory-building procedure and has not been empirically validated. The framework organizes the recurrent findings across clinical, institutional, and regulatory levels and should be evaluated prospectively before use as a normative standard.)

Policy Variability and Global Ethical Gaps

The umbrella review revealed significant variability in AI-related nursing governance policies across different countries and healthcare systems. Reviews pointed out that some high-income countries have started to develop regulatory frameworks that address AI ethics, transparency, and accountability, while others lag significantly behind (Karimian et al., 2022; Martinez-Ortigosa et al., 2023). This uneven landscape complicates efforts to standardize nursing practices, particularly in globalized workforces where nurses often move across borders or participate in transnational health initiatives (Roppelt et al., 2024).

A particular concern was the lack of global consensus on ethical priorities in AI governance for nursing (Joo et al., 2025). While some nations emphasize data protection and patient rights, others focus more heavily on innovation and cost-efficiency (Tun et al., 2025). This imbalance creates ethical tension and regulatory confusion, especially in multinational institutions or collaborative research networks (Ruksakulpiwat et al., 2024). The reviews called for international collaboration among nursing boards, policy bodies, and professional associations to develop shared ethical guidelines and harmonized regulatory approaches (Ibrahim et al., 2025; Mikkonen et al., 2025). Global alignment, while challenging, is essential for maintaining the ethical integrity of the nursing profession in an increasingly AI-driven world.

Education, Competency, and Ethical Preparedness

The need for education and competency-building in AI ethics emerged as a recurrent concern across the included reviews. Nursing education programs have not kept pace with the rapid expansion of AI technologies in healthcare, creating gaps in knowledge and preparedness among both students and practicing nurses (Alnawafleh et al., 2025). Ethical preparedness, defined as the knowledge, competencies, critical awareness, and ethical decision-making capacity required to responsibly evaluate, govern, and use AI technologies, was consistently identified as an essential component of modern nursing practice (Alqaissi & Qtait, 2025). In the digital age, such preparedness requires not only technical understanding of AI systems but also the ability to critically assess their implications for patient care, professional accountability, and ethical decision-making (Badawy et al., 2025).

Several reviews emphasized the importance of integrating AI ethics into undergraduate and postgraduate nursing curricula, continuing professional development programs, and licensure requirements to strengthen workforce readiness (Joo et al., 2025; Mohammed et al., 2025). Beyond formal education, experiential learning and interdisciplinary collaboration were recommended as strategies for developing ethical reflexivity and critical thinking regarding AI implementation (Ibrahim et al., 2025). Nursing students and professionals should be encouraged to engage in issues such as data justice, algorithmic bias, and the broader sociopolitical implications of AI adoption in healthcare (Gonzalez-Garcia et al., 2024). The evidence suggests that cultivating an ethical mindset is essential for ensuring that AI technologies are used responsibly and remain aligned with the core nursing values of care, compassion, equity, and patient advocacy. Without deliberate investment in ethical capacity-building, nurses may become technologically proficient yet insufficiently prepared to address the ethical and governance challenges associated with AI integration (Al Khatib & Ndiaye, 2025).

Convergence, Divergence, and Strength of Evidence Across Reviews

Across the included reviews, strong convergence was observed regarding the ethical and governance challenges associated with AI in nursing. Concerns related to algorithmic bias, data privacy, transparency, professional accountability, and the need for stronger AI governance were consistently reported (Mohammed et al., 2025). Reviews also agreed that current regulatory frameworks have not kept pace with rapid AI implementation and that nurses remain underrepresented in AI governance and policymaking processes (Burford et al., 2025; Roppelt et al., 2024). Similarly, AI literacy, ethics education, and leadership development were widely identified as priorities for responsible implementation (Von Gerich et al., 2022). At a higher level, the themes formed an accountability chain linking data governance, bias and safety, transparency, institutional authority, and professional responsibility. Although global overlap was slight (CCA 0.54%), localized overlap meant that agreement between reviews could not automatically be interpreted as independent confirmation. Repeated primary studies were therefore not given additional weight in the narrative synthesis.

Despite this agreement, some divergence was evident. While several reviews highlighted AI’s potential to enhance clinical decision-making and professional autonomy (Pereira et al., 2025; Tun et al., 2025), others warned that reliance on algorithmic recommendations could reduce clinical judgment and increase dependency on automated systems (Labrague & Al Harrasi, 2025; Mikkonen et al., 2025). Differences in findings appeared to reflect variations in AI applications, healthcare settings, and implementation contexts. Overall, the most consistently supported evidence emphasized accountability, transparency, algorithmic bias, data governance, and greater nursing involvement in AI governance as critical priorities for future policy and practice.

Discussion

AI can improve decisions and workflow while challenging autonomy and accountability. Governance models should respond as algorithms increasingly influence clinical and organizational processes (Prakash et al., 2022), because human-centered ethical frameworks may not address opacity and automated governance (Hou et al., 2025; Ratti et al., 2025; Yu & Zhai, 2024). Risks also differ by application: predictive analytics emphasize fairness (Karimian et al., 2022), decision support raises explainability and accountability (O’Connor et al., 2023), and generative AI adds misinformation and privacy risks (Mohammad Amini et al., 2023; Yu & Zhai, 2024). Biased data can reproduce inequity in prioritization and allocation (Pailaha, 2023), conflicting with nursing commitments to justice and inclusion (Labrague et al., 2023; Yang et al., 2025). Weak auditing and monitoring indicate limited regulatory preparedness (Mennella et al., 2024), making fairness-focused governance a priority (Benzinger et al., 2023). Extensive data reuse also challenges one-time consent. Informational asymmetry limits meaningful understanding (Lambert et al., 2023); policies should therefore address ownership, secondary use, privacy, and surveillance (Higgins et al., 2023). Unclear responsibility when AI conflicts with judgment can create moral distress and legal vulnerability (Gerlach et al., 2025; Saif-Ur-Rahman et al., 2023). Regulation should preserve nurses’ judgment while preventing inappropriate liability (Johnson et al., 2024), supported by updated competencies and professional guidance (Wangpitipanit et al., 2024).

Nurses encounter AI across monitoring, documentation, communication, prioritization, and workforce management, alongside advocacy and coordination (Mohammed et al., 2025). AI can therefore reshape professional identity and therapeutic relationships (Arcadi, 2025), requiring nursing-specific attention to autonomy, advocacy, accountability, and regulation (Van Der Gaag et al., 2023). Limited nursing involvement in efficiency-driven adoption weakens oversight (Göktepe & Sarıköse, 2025). Nurses should participate in design, deployment, and evaluation (Rony et al., 2025), while education and interdisciplinary governance strengthen accountability (Namdar Areshtanab et al., 2025; Sengul et al., 2025).

The synthesis also suggests that governance should be understood as a continuing lifecycle rather than a one-time approval decision (Cao et al., 2025). Pre-implementation review should examine intended use, training data, subgroup performance, workflow fit, and responsibility for foreseeable errors. During implementation, organizations should document nurses’ authority to override recommendations, establish escalation routes, and monitor changes in workload, communication, and patient participation (Mikkonen et al., 2025; Pereira et al., 2025). Post-deployment oversight should include incident reporting, periodic bias audits, model-drift assessment, and review of effects on autonomy and care relationships (Ruksakulpiwat et al., 2024). Regulators can define minimum expectations; institutions can translate them into procedures; developers can provide documentation and technical monitoring; and nurses can contribute contextual knowledge (Mohammed et al., 2025; Van Der Gaag et al., 2023). This lifecycle interpretation shows how transparency, fairness, accountability, data stewardship, and nursing leadership can operate as mutually reinforcing governance controls.

Strengths and Limitations

This review synthesized AI-related nursing governance across settings and jurisdictions using JBI appraisal and PRISMA reporting. Included reviews varied in design, scope, and quality. Formal overlap assessment was possible for 23 reviews and indicated slight overlap (CCA 0.54%); however, eight reviews were not represented because complete, consistent, or separable primary-study lists were unavailable. Their exclusion may underestimate overlap across all 31 reviews, and inconsistent citations or missing DOIs may have caused residual matching error. English-language restriction, no eligible grey literature, and heterogeneous terminology may also have limited comprehensiveness.

Implications for Practice

Healthcare organizations, nursing leaders, educators, and policymakers should develop transparent governance, ethical guidance, workforce training, and regulatory oversight for safe and equitable AI integration. These measures may strengthen trust, patient safety, accountability, and sustainability. In practice, organizations should pilot AI systems and monitor safety, fairness, usability, workload, and patient experience. Nursing representatives should participate in procurement, validation, ethics review, incident investigation, and reauthorization. Regulators and educators should align competencies so nurses can recognize unreliable outputs, document concerns, communicate uncertainty, and escalate without compromising accountability (Table 2).

Table 2.

Policy and Practice Recommendations for AI Governance in Nursing

Level	Governance action	Responsible stakeholders	Evidence status
Regulatory	Define minimum standards for safety, fairness, transparency, data stewardship, human oversight, and professional liability.	Regulators; professional boards	Evidence-informed; prospective evaluation needed
Institutional	Establish multidisciplinary oversight with nursing representation; audit subgroup performance; define override, escalation, and incident-reporting processes.	Healthcare organizations; AI developers; nursing leaders	Consistently recommended; limited evaluated implementation
Clinical	Preserve professional judgment, document disagreement with AI outputs, communicate uncertainty, and protect patient consent and advocacy.	Nurses; clinical leaders	Strong conceptual convergence; limited outcome evaluation
Education	Integrate AI appraisal, privacy, bias, accountability, and governance into curricula, continuing development, and competency standards.	Educators; accreditors; regulators	Consistently recommended; effectiveness uncertain

Conclusion

AI is reshaping nursing governance by challenging ethical norms, regulation, and professional autonomy. Its benefits in efficiency and decision support coexist with opacity, bias, and unclear accountability. Policy responses should involve nurses, protect patient advocacy, address inequality, support ethical preparedness, participatory oversight, and regulatory alignment so that AI strengthens nursing practice and preserves the integrity of care. Future governance should move from broad ethical endorsement toward auditable responsibilities across the AI lifecycle. Evidence remains insufficient to identify one superior regulatory model, but the review supports clearer role allocation, nursing representation, transparent monitoring, and protection of professional judgment. Prospective evaluation should determine whether these measures improve safety, equity, accountability, and patient-centered care across nursing contexts.

ORCID iDs

Daifallah M. Alrazeeni https://orcid.org/0000-0002-8149-8650

Moustaq Karim Khan Rony https://orcid.org/0000-0002-6905-0554

Data Availability Statement

The review-level extraction, study-characteristics, thematic-coding, and methodological-appraisal data are provided in the supplementary files. No new primary-study data were generated.*