Deepfake: what it is and why it matters to the clinic

· 2 min read

Quick facts

Definition
Synthetic image, video, or audio that convincingly imitates a real person, generated by AI.
Risk
Fraud, false identity, defamation, and use of a person's image without authorization.
Difference
An illustrative procedure simulation is a planning tool, with no intent to deceive.
Caution
Identify the image as a simulation and never present it as an actual result.

A deepfake is synthetic content, video, image, or audio, produced by artificial intelligence to convincingly imitate a real person. The term is usually associated with fraud, scams, defamation, and use of a face without authorization. Generative AI models have made creation easier and more realistic.

Why does this matter to the clinic?

There are three fronts:

  • Image of the professional or the clinic: fake profiles with the face and voice of the physician, dentist, or biomedical professional to sell products or procedures.
  • Image of the patient: face photos copied to create manipulated images or fake profiles. Hence the care with storage and access. See facial biometric data.
  • Marketing: use of AI-generated images as if they were real before and after photos, which can amount to deceptive advertising.

Is a procedure simulation a deepfake?

Technically, a simulation generated by AI from the patient's photo uses the same family of technology. The difference lies in purpose and transparency: an illustrative simulation serves planning, is shown to the patient themselves, and is identified as a projection. The problem arises when someone presents it as a real result.

Use Nature
Showing it to the patient, with a simulation notice Planning tool
Posting it as if it were an actual result Consumer deception
Using a third party's face without authorization Violation of image rights and of the LGPD

Frequently asked questions

What is a deepfake?

It is synthetic media created with artificial intelligence to make it look like a person said or did something that did not happen. It can be video, a photo, or a voice.

Is a before and after projection a deepfake?

Technically it uses generative AI on the patient's photo, but the goal is planning, identified as a simulation. If it is published as an actual result, it becomes deceptive.

Terms mentioned