Can a clinic use AI-generated images and writing on its website?

Short answer: assess the published claim, not just the tool

In South Korea, using AI to draft text or create a background image is not, by itself, a blanket violation of medical advertising rules. The Medical Service Act does prohibit false or exaggerated advertising and claims that mislead people about treatment effects. Calling a fictional clinician, room, device, or outcome ‘ours’ is not made accurate by saying AI was used.

Where can a page depart from reality?

Imagine using a generated room image as if it shows your clinic, or introducing a fictional person as a treating clinician. Do not present a generated before-and-after image as a real treatment result. An abstract image used for atmosphere should have a context and description that do not imply it depicts an actual room or outcome.

Retouching a real director's or clinician's photo, or turning it into an illustration, can be considered as a presentation choice. Have the clinic approve the final image, ensure the person and qualifications cannot be mistaken, and label an illustration if it could look like a real photo.

The same check applies to writing. AI may confidently add a service the clinic does not offer, an unverified device, or an unsupported recovery time. Do not publish those statements without verification.

Create a review path for copy and images

Start with clinic-approved facts about its name, address, hours, scope of care, and clinician qualifications. The relevant clinician checks descriptions of care, effects, and risks. The operations lead checks whether people, rooms, and equipment in images match reality.

• Keep the source facts, AI draft, and clinic-approved version distinct.
• If you edit a real photo, do not make the outcome or shooting conditions look materially different.
• Do not send patient names, contact details, records, or identifiable photos into the content production workflow.

What happens after publication?

Changes to hours or clinicians also require another look at AI-drafted introductions. Record who approved each page and when; agree on when a clinician will recheck care-related wording. This workflow manages accuracy. It does not certify every possible AI use as lawful.