Ethics of Artificial Intelligence In Social Work Practice, Research and Education: A Scoping Review

Background: Artificial intelligence (AI) is increasingly integrated across multiple domains of social work, including professional practice, research, and education. AI-powered tools are being utilized in various areas of social work, such as enhancing client engagement, supporting decision-making processes, assisting in research design and data analysis, and providing personalized learning experiences for students. Social workers have responded by raising concerns about the ethics of AI in social work practice, research and education, including issues related to algorithmic bias, predictive analytics, informed consent, and professional competence. Given the siloed nature of social work and the difficulty of applying scholarship from other fields, the limited literature on social work research ethics, the profession’s recent interest in AI, and the central role of ethics in future AI applications a scoping review is needed to identify, map, and critically examine how ethical considerations are currently conceptualized in the literature.

Methods: This scoping review followed the methodology proposed by Arksey and O’Malley (2005) and the PRISMA-ScR guidelines (Tricco et al., 2018). The guiding question for this review was: What are the ethical considerations of using artificial intelligence in social work practice, research, and education? To conduct a comprehensive search, the following databases were used: Academic Search Complete, ERIC, Communication & Mass Media Complete, PsycARTICLES, PsycINFO, CINAHL, Web of Science, Social Work Abstracts, Social Services Abstracts, and PubMed.

Results: The initial search yielded 431 studies. A total of 205 duplicates were identified and removed using Covidence. After screening titles and abstracts, 21 articles remained reflecting a geographically diverse sample, although some articles (n=9) did not specify the study location. Some studies discussed the use of a variety of generative AI products, primarily ChatGPT, but also including Gemini, Copilot, Claude, Perplexity and pilot tools created for caseworkers. The majority of practice articles focused on ethical standards regarding social workers’ responsibilities to clients (n=17), including privacy and confidentiality (n=13), informed consent (n=6) and competence (n=6). Additionally, research-focused articles addressed the need to maintain an anti-racist lens and test AI models for bias (n=7). Education-focused articles identified the need to balance AI-related educational competencies, while honoring student and instructor agency. Notably, the fewest number of articles discussed ethical concerns related to dignity and worth of the individual (n=4).

Conclusion and Implications: Findings highlighted both concerns and potential benefits across the 21 articles examined in this scoping review. All authors highlighted concerns about social justice implications of bias in AI and its black-box nature. This review indicates that an ethical approach to AI in social work requires not only individual judgment but also interprofessional regulatory guardrails that establish clear standards for acceptable use. Each major organization has taken steps towards this goal, including the proposed NASW AI Task Force, the SSWR ETHICS guidelines (Lee et al., 2026) and the proposed AI competency for the 2029 EPAS. The rapid evolution of AI and absence of accountability structures leaves practitioners, researchers, and educators navigating complex risks, indicating the need for urgent ethical guidelines for AI use within social work.