By Cody Saunders, LMSW | October 15, 2026
AI raises new questions about familiar social work values. A tool may save typing, yet change whose voice is heard. It may find patterns, yet miss the reason behind them. The key question is how its use affects people, their choices, and the trust at the heart of care.
Two 2023 resources offer a useful starting point: NASW’s 8 Ethical Considerations for the Use of Artificial Intelligence in Social Work and Frederic G. Reamer’s Artificial Intelligence in Social Work: Exploring Key Ethical Considerations, prepared for a July 2023 NASW-sponsored ethics roundtable. This article explains their main ideas in plain language and uses made-up examples. It is general education, not a clinical guide, consent form, or implementation plan.
Read these sources in context
The tip sheet connects AI use to existing social work ethics. Reamer’s slides raise questions about possible uses and harms. Neither document proves that a product works or is safe. Their claims about the state of research describe 2023; they are not a current review of all AI evidence.
The NASW Code of Ethics sets professional ethical standards. It is different from a statute, a licensing rule, or an employer’s policy. How each applies depends on the role and setting. A professional recommendation about consent does not settle every legal question about data sharing. This website is independent and is not endorsed by NASW.
Skill means more than knowing which button to press
NASW’s tip sheet stresses competence: having the knowledge and skill needed for the work. With AI, this includes knowing the tool’s limits and recognizing when its answer needs more scrutiny. It also includes keeping up with changes. A product update can add AI to software a team already uses. Familiar software may now handle information in a new way. NASW tip sheet, points 1, 6, and 7.
Consider an AI draft that sounds clear but leaves out a client’s stated concern. Good grammar cannot show whether the draft is complete. Understanding the source and the task matters. Learning about AI supports judgment; it does not make every use suitable for care.
Client choice is part of the picture
The tip sheet links disclosure and informed consent to client self-determination. That means a person has a meaningful role in identifying their goals and making choices about services. The Code’s technology provisions address informed consent and a person’s ability to use and benefit from the technology. Code, sections 1.02 and 1.03.
Imagine a client whose main goal is to return to work. A tool instead makes symptom scores the center of a draft plan. Even if the scores are accurate, the draft may miss what matters most to that person. AI can support a person’s goals without defining those goals for them.
Knowing that AI is involved is different from understanding what it does. A broad label such as “digital support” may say little about recording, data access, or the role of a generated answer. Specific consent and disclosure practices require qualified review of the setting. This article does not supply a consent script or form.
Privacy includes what happens after collection
Reamer’s slides ask how information may be reused, who controls it, and whether it could support unwanted monitoring. The tip sheet also highlights confidentiality. These are questions about the full life of data, not just whether a file has a password. Roundtable slides, page 4; NASW’s event information.
A mood app may collect entries each day. Those entries could reveal a pattern far beyond one visit. Who may receive that pattern, and for what purpose, affects privacy and personal control. This example does not establish how a particular app handles data. See Can I Put Client Information Into an AI Tool? for the difference between privacy concerns and a specific legal determination.
Fairness includes access and whose experience counts
The tip sheet connects bias to discrimination and cultural competence. A tool may perform unevenly if the people behind its data do not reflect those using it. Language, income, disability, culture, and access to devices can also shape who can use a service. NASW tip sheet, point 4.
For example, an app-only service may be hard to use without a private phone or stable internet. That barrier exists even if the app gives accurate answers. A high average score also may hide poorer results for a smaller group. Evidence about the relevant population matters alongside overall performance.
Availability is different from continuity of care
A chatbot can produce text at any hour without offering reliable clinical support. Reamer’s slides raise concerns about wrong clinical conclusions and gaps in timely help. These are risks to consider, not proof that every AI tool causes harm. Roundtable slides, page 4.
A person may feel attached to a tool that later changes, loses conversation history, or stops working. The relationship and its limits matter as well as the words on the screen. This is an educational concern about reliance and continuity, not a crisis-response procedure or treatment recommendation.
Honesty still matters when AI writes
Reamer also raises plagiarism and misrepresentation. A generated report may repeat material without clear credit, invent a source, or describe work that never took place. Fluent writing is not evidence that the facts or sources are sound. Roundtable slides, page 4.
A grant draft that invents a program result is not made accurate by a human name at the bottom. Authorship, source credit, and truthfulness remain distinct questions. Copyright and ownership questions need their own review; an AI label does not settle them.
Governance connects these concerns
The tip sheet discusses technology policies. Reamer’s slides describe governance ideas such as diverse input, review of results, and attention to bias over time. These are professional discussion points, not a required ten-step process or a validated safety test. NASW tip sheet, point 8; roundtable slides, page 8.
They fit within this site’s existing five domains: Clinical Appropriateness; Privacy & Security; Reliability & Safety; Human Oversight; and Governance & Accountability. They do not create a second framework. Staff and client perspectives can reveal problems that a product demonstration misses. See the master educational framework.
Keep the purpose in view
AI may ease some administrative work. It may also add checking, repair, and new costs. Ethical discussion helps make those tradeoffs visible without promising better care or savings. The aim is thoughtful use of evidence and professional judgment, with people’s needs and choices in view.
Sources and scope
- NASW, 8 Ethical Considerations for the Use of Artificial Intelligence in Social Work. Supplied 2023 tip sheet, two pages. NASW’s source page may require sign-in. Professional guidance, not legislation.
- NASW Ethics Table Talks. Source event information. This article draws on the supplied nine-page presentation by Frederic G. Reamer, July 2023, especially pages 4–8; it does not claim to summarize a recording.
- NASW Code of Ethics: Ethical Responsibilities to Clients. Sections 1.02–1.05 and 1.07 provide related context. The current source page was checked October 15, 2026.
- NASW AI and Social Work resource hub. A place to find newer material. Links to a professional organization do not imply endorsement of this website.
These sources do not establish product safety, clinical effectiveness, HIPAA compliance, or permission for a particular data use. Specific decisions belong to qualified professionals and the responsible organization.