Artificial Intelligence In U.S. Social Work Practice: Types of Use, Attitudes, and Ethical Concerns

Background: Health and human services staff are increasingly using AI tools, yet there is little evidence on how social workers are using AI, how they perceive its risks and benefits, and what training and other supports are needed for ethical implementation. This study examines current AI use, attitudes, organizational readiness, and ethical concerns among U.S.-based social workers. Methods: Our research team employed a cross-sectional, mixed-methods survey design through a national online survey between October 2025 and February 2026 that was distributed by the National Association of Social Workers, with additional distribution within the Department of Veteran Affairs and the Department of Defense. A total of 1,179 U.S. social workers participated, representing diverse practice settings and career stages. We conducted quantitative and qualitative analyses to assess patterns of AI use, perceived usefulness, barriers, and support needs. Results: Nearly two-thirds of respondents (63.5%) reported current AI use, most commonly for correspondence, documentation, administrative tasks, and information seeking. AI was viewed as most useful for reducing administrative burden rather than for use in clinical practice. More than 58% of respondents from across 18 social work practice indicate use of AI with the highest rates of use (over 80%) from Policy, Academia and Professional Development. High levels of use were also reported in direct service practice areas, including advocacy and community organizing (75.0%), developmental disabilities (73.3%) and justice and corrections (70.4%). Nearly half of respondents (44.7%) report using ChatGPT or other OpenAI products, with other platforms being used at notably lower rates. While most respondents (73%) anticipated AI playing a larger role in the future of social work practice, attitudes reflected cautious optimism more than uncritical adoption. Ethical concerns were prevalent, particularly regarding data privacy and security, client safety, bias, environmental impact, lack of regulation, and erosion of professional skills. Thematic analysis of open-ended responses revealed five dominant themes: ethical concerns, general resistance to AI, bounded acceptance focused on task-specific use, the irreplaceability of human judgment and relationships, and an urgent need for professional guidance and regulation. As to extent of implementation in their organizations, AI adoption remains limited, with minimal adoption within departments (34.2%) and organizations (32.6%). Implications: Findings highlight significant engagement with AI alongside ethical concern and implementation uncertainty. Results underscore the need for social work-specific ethical guidelines, training, regulation, and practitioner leadership to ensure AI enhances, rather than undermines, the profession’s core values and commitment to human-centered practice. Many additional areas of research are needed: 1. research to understand how clients perceive AI involvement in their care, particularly with regard to trust, safety, transparency, and cultural responsiveness, is essential for ethical implementation and informed consent, 2. practice-area–specific studies that more deeply examine how AI is being used within particular contexts, such as mental health, child welfare, healthcare, schools, and justice settings, and 3. evaluation and effectiveness research examining whether and under what conditions AI tools improve outcomes for social workers and the communities they serve.