Using generative AI safely
About 25 minutes · plain language
Generative AI can save time when you study, teach or do research, if you use it with clear limits. This lesson gives you a simple four-step routine and clear rules about what you must never put into a public AI tool.
- Write a clear prompt and check the answer against sources.
- Decide what must never be pasted into a public AI tool.
Key terms
- Prompt
- The instruction or question you give to a generative AI tool.Example: “Explain fluid balance in 150 words for a first-year nursing student.”
- Confidentiality
- The duty to keep personal and unpublished information private.Example: Patient details, exam papers and manuscripts sent to you for review.
- Identifiable data
- Information that could reveal who a person is, alone or combined with other details.Example: Age, ward, date and diagnosis together can identify a patient even without a name.
- Verification
- Checking AI output against a trusted source before you use it.Example: Checking a drug dose in the national formulary.
- Disclosure
- Saying openly that, and how, you used AI.Example: A sentence in the methods of a paper naming the tool and its purpose.
- AI-use log
- A simple record of when and how you used AI.Example: Date, tool, version, purpose, how you checked the output.
Step 1: Ask clearly
Tell the tool who you are, what you need, the format, and the limits. For example: “I am a nursing student. Explain fluid balance in 150 words for a first-year exam. Use simple language. Do not include references.” A clear prompt gives a more useful answer, but it does not make the answer true.
Step 2: Check every fact
Check numbers, doses, definitions and claims against a trusted source: your textbook, a clinical guideline or the formulary. Look up every reference in PubMed or the journal website, because invented references are common (Lesson 1). Never use AI output for a clinical decision without checking it against guidelines and your own assessment.
Step 3: Record what you used
Keep a short AI-use log: date, tool, version, what you asked, and how you checked the answer. It takes a minute and makes disclosure easy later.
Step 4: Disclose
Many universities and almost all major journals now ask you to disclose substantial use of AI. AI cannot be listed as an author, and you remain responsible for everything you submit. In research, the GAMER guideline lists what to report: which tool and version, what it was used for, and how its output was checked.
What you must never paste in
Text you type into a public AI tool may be stored or used by the company that runs it. Never paste identifiable patient information, even without a name. Do not paste exam papers or manuscripts you have been asked to review; ICMJE notes that confidentiality may rule out uploading a manuscript to AI tools that cannot assure it. Ask first before using your own unpublished data or institutional documents. The figure shows these as traffic lights.
This is not only good practice; it is often the law. Data-protection laws, such as the EU General Data Protection Regulation and Oman’s Personal Data Protection Law, treat health data as specially protected and limit where they may be sent, including to AI services hosted in other countries (Lesson 7).
Case: practice questions for a pharmacology class
A nurse educator asks a chatbot to draft ten multiple-choice questions on anticoagulants. She checks each answer against the course textbook and finds two errors: one wrong antidote and one outdated monitoring target. She corrects them, records the tool and date in her log, and adds a line to the course page: “AI helped draft these questions; all answers were checked by the teaching team.”
Common mistakes
- Copying references without checking them.
- Pasting case details “without the name” that still identify the patient.
- Forgetting to disclose.
Check your understanding
1. A colleague pastes “72-year-old man, Ward 4, admitted yesterday with a stroke” into a public chatbot. Is this acceptable?
No. Even without a name, these details together could identify the patient.
2. What four steps should you follow every time you use generative AI?
Ask clearly, check every fact, record what you used, and disclose.
3. Can ChatGPT be listed as a co-author on a paper?
No. AI cannot be an author; authors stay responsible and disclose how AI was used.
Try it
- Write a prompt for a task you do often, using “who, what, format, limits”.
- Start an AI-use log today: date, tool, version, purpose, how you checked.
Remember
- Ask clearly, check every fact, record, disclose.
- Never paste identifiable patient data into a public AI tool.
- You stay responsible for anything AI helped you produce.
- World Health Organization. Ethics and governance of artificial intelligence for health: guidance on large multi-modal models. Geneva: WHO; 2024. www.who.int
- International Committee of Medical Journal Editors. Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. Updated January 2026 (Section V: Use of artificial intelligence). www.icmje.org
- Luo X, Tham YC, Giuffre M, et al. Reporting guideline for the use of Generative Artificial intelligence tools in MEdical Research: the GAMER Statement. BMJ Evid Based Med. 2025;30(6):390-400. doi.org
- Miao F, Holmes W. Guidance for generative AI in education and research. Paris: UNESCO; 2023. doi.org
- Committee on Publication Ethics. Authorship and AI tools: COPE position statement. COPE; 2023. publicationethics.org
- Walters WH, Wilder EI. Fabrication and errors in the bibliographic citations generated by ChatGPT. Sci Rep. 2023;13(1):14045. pubmed.ncbi.nlm.nih.gov
- Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data (General Data Protection Regulation). Official Journal of the European Union. 2016;L119:1-88. eur-lex.europa.eu
- Sultanate of Oman. Royal Decree No. 6/2022 promulgating the Personal Data Protection Law. Muscat; 2022. decree.om
Lesson text is our own plain-language summary of these sources; it is not endorsed by their authors. Check the original before relying on any detail.