AI for Doctors and Clinics in Oman: Documentation, Patient Secrets and Safe Daily Use

Doctors and clinics in Oman can use AI every day for the work around the consultation:

• drafting clinical notes from dictation;
• referral letters and discharge summaries;
• patient instructions in Arabic and English;
• appointment and insurance paperwork;
• summaries of guidelines for clinicians.

Every clinical output is a draft that the treating practitioner reviews and owns.

Three legal points shape the rest:

1. Documentation. Oman’s medical practice law requires practitioners to work with accuracy according to scientific and technical principles, and to document the patient’s condition, diagnosis and treatment.
2. Patient secrets. The same law forbids disclosing patient secrets without the patient’s written consent, subject to listed exceptions.
3. Health data. Health data is sensitive under the Personal Data Protection Law: processing it requires a permit from the Ministry. Since September 2026, the definition also covers data about the health care services a person received.

So AI assists and the clinician decides. Patient data goes only into tools approved for it, under contract. Patients are told when AI listens or writes. Nothing from AI reaches a patient unreviewed.

The short answer

• Best uses: documentation drafts, letters and summaries, patient instructions, administrative paperwork, and guideline summaries for clinicians.
• Not for: unreviewed diagnosis or treatment decisions, or medical advice sent straight to patients.
• Law: Articles 10 and 33 of the medical practice law issued by Royal Decree 75/2019; the Article 5 permit for health data under the PDPL; the wider definition of health data since 7 September 2026.
• Risks named by WHO: “false, inaccurate, biased, or incomplete statements”, and “automation bias”.
• Controls: approved tools, processor contracts, minimum data, patient information, clinician sign-off and logs.

What WHO says about large AI models in health

WHO’s January 2024 guidance on large multi-modal models lists five broad applications in health:

1. diagnosis and clinical care, such as responding to patients’ written queries;
2. patient-guided use, such as investigating symptoms and treatment;
3. clerical and administrative tasks, such as documenting and summarising patient visits within electronic health records;
4. medical and nursing education;
5. scientific research and drug development.

It also records “documented risks of producing false, inaccurate, biased, or incomplete statements”. And it warns that these models can encourage “‘automation bias’ by health care professionals and patients”, meaning errors slip through because the output looks authoritative.

For a clinic, the clerical and administrative applications are where the benefit is clearest and the risk easiest to control.

Where AI helps in a clinic

Task	How AI helps	Who checks what
Clinical notes	Drafts a structured note from dictation or a recorded consultation	The treating clinician reads, corrects and signs it
Referral letters and discharge summaries	Drafts from the record and the clinician’s notes	The clinician checks diagnoses, medications and follow-up
Patient instructions	Turns approved instructions into clear Arabic and English at the right reading level	A clinician approves each template before use
Administrative paperwork	Drafts insurance forms and appointment letters from record fields	Staff check every field against the record
Guideline and literature summaries	Summarises a guideline or paper for a clinician, with references	The clinician reads the source before applying it
Front-desk messages	Answers questions on hours, location, preparation and booking	Anything clinical goes to staff; the bot never advises

Booking and reminders are covered in booking and appointment systems.

What Omani law requires

The practitioner’s duties. The Law Governing the Practice of the Medical Profession and Allied Health Professions, issued by Royal Decree 75/2019, sets out practitioners’ duties in Article 10. They include performing their work with the accuracy and honesty the profession demands, according to scientific and technical principles. They also include documenting the patient’s health condition, diagnosis and prescribed treatment in the designated record. An AI draft does not change who is responsible for the note. The practitioner who signs it is.

Patient secrets. Article 33 prohibits practitioners from disclosing patient secrets learned through their practice without the patient’s written consent. The exceptions are listed, among them:

• disclosure to competent authorities to prevent or detect a crime;
• cases of diseases that threaten public health;
• hereditary diseases;
• the practitioner’s own defence.

Treat any AI vendor that receives patient information as an outside party with access to those secrets. That means a written contract, the minimum data needed, and advice on whether patient consent is required for your setup.

Health data under the PDPL. Article 5 of the Personal Data Protection Law prohibits processing health data without a permit from the Ministry. Royal Decree 68/2026 widened the definition. Since 7 September 2026, health data also covers data relating to the provision of health care services that reveals a person’s health status, so appointment histories and insurance claims count too. Breaching Article 5 carries a fine of OMR 15,000 to 20,000. What Royal Decree 68/2026 changed covers the amendments.

Data leaving Oman. If the AI tool runs on servers abroad, sending patient data is a cross-border transfer. It needs explicit consent before the transfer, unless the data is anonymised or an international agreement applies. That makes in-country or self-hosted options worth weighing for clinical documentation. Where your company’s data is allowed to live covers Oman’s hosting rules.

AI scribes: recording the consultation

Ambient “AI scribes” record the consultation and draft the note. They can save clinicians time, but they record some of the most private conversations a person has. Before using one:

• Tell the patient before recording, in plain language, and offer to switch it off. Record the patient’s choice.
• Choose the tool on its data terms: where audio and transcripts are processed and stored, how long they are kept, and whether the provider may use them.
• Delete the audio once the note is signed, unless there is a clear reason to keep it.
• Check the note against the consultation, especially negatives such as “no chest pain”, medication names, doses and which side of the body.
• Test Arabic and mixed-language consultations, including dialect and switching between Arabic and English, before relying on the tool.

Recording conversations without people’s knowledge can also engage Oman’s Cybercrime Law. AI meeting assistants and recording consent explains the rules.

What not to use AI for

• Diagnosis or treatment decisions without the clinician’s own assessment. Software that diagnoses or recommends treatment is regulated as a medical device in many countries. Check with the Ministry of Health before deploying anything of that kind.
• Medical advice sent directly to patients by a chatbot.
• Patient data in public AI tools.
• Unreviewed translations of clinical instructions.

A safe starting plan for a clinic

1. Pick one clerical task, such as referral letters or patient instruction templates.
2. Approve the tool: business terms, data location, a processor contract, and how it fits your health data permit.
3. Write the prompts with the clinicians who will use them, and keep them in a shared prompt library.
4. Run a supervised pilot for a month, with every output checked and every correction logged.
5. Decide with evidence whether to extend it. Is AI actually saving your business time? covers how to measure.

---

Frequently asked questions

Can doctors in Oman use AI to write clinical notes? Yes, as a drafting aid. Under the medical practice law the practitioner remains responsible for documenting the patient’s condition, diagnosis and treatment, so the treating clinician must read, correct and sign every AI draft.

Do patients have to consent to AI scribes? Tell patients before any recording and offer to switch it off. Whether written consent is legally required depends on your setup, including where the data is processed; medical secrecy, data protection and cybercrime rules all bear on it. Take advice, and record each patient’s choice.

Is patient data allowed in AI tools? Only in tools your clinic has approved for health data, under a written contract, with the minimum data necessary. Health data is sensitive under the PDPL and needs a Ministry permit to process, and transfers abroad carry their own conditions.

Can a clinic chatbot give medical advice? No. A front-desk chatbot can handle opening hours, locations, preparation instructions and bookings, and must hand anything clinical to staff.

What does WHO say about AI in health care? WHO’s 2024 guidance on large multi-modal models lists uses from clinical care to administration. It warns of false, inaccurate, biased or incomplete statements, and of “automation bias”, where professionals overlook errors in automated output.

Did the 2026 amendments change anything for clinics? Yes. Since 7 September 2026, health data includes data about the provision of health care services that reveals a person’s health status, so appointment histories and insurance claims are covered. The amendments also added a right to object to automated decisions, with human review.

---

This article states the position under the laws in force on 28 October 2026, the date of publication. It is not legal or medical advice.

Sources: Royal Decree 75/2019 — Law Governing the Practice of the Medical Profession and Allied Health Professions · Royal Decree 68/2026 amending the Personal Data Protection Law · Royal Decree 6/2022 — Personal Data Protection Law (Arabic text) · WHO — Guidance on large multi-modal models in health

Want AI that saves clinicians time without putting patient secrets at risk? ZAWAT builds AI and automation and business systems for clinics, with data controls, approvals and audit trails designed in. Book a call.