How to Talk to Your Patients About AI Scribes: Scripts and FAQs
A clinician who introduces an AI scribe poorly can make a patient uncomfortable in the first 30 seconds of a session. A clinician who introduces it well gets a nod, a "that sounds great," and a session that proceeds exactly as it would have without the scribe — except the note is already done at the end. The difference is almost entirely in how the conversation is framed. Here are scripts, answers to common patient questions, and the consent best practices that make the introduction feel natural rather than intrusive.
The 30-Second Introduction Script
You don't need a long preamble. Most patients respond well to a brief, confident explanation that focuses on the benefit to them — you being more present — rather than the technology itself.
Script:
"I use an AI tool that listens to our session and helps me write my notes afterward. It means I can be fully focused on you during our time together instead of typing. The recording is used only to generate my clinical note, and it's deleted afterward. Is that okay with you?"
Three things make this work:
- It's short. Patients don't want a technical explanation.
- It leads with the patient benefit (your full attention), not the tool.
- It ends with a question, which makes consent a conversation rather than a form.
Handling Patient Questions
"Are you recording me?"
"Yes, the session audio is captured so the AI can transcribe it and help me write my note. The recording is used only for that purpose and is deleted after the note is generated. It's not shared with anyone else, and you won't be identified by name in any stored data."
"Where does the recording go?"
"The audio is processed by a HIPAA-compliant service that transcribes it and generates a draft note for me to review. The company that processes it has signed a Business Associate Agreement, which is the legal document that makes them responsible for keeping the data secure under HIPAA. The audio is deleted after the note is generated."
"Can I say no?"
"Absolutely. If you'd prefer I take notes the traditional way, I'm happy to do that. It won't change anything about our session."
This is important: the answer should be a genuine yes, not a reluctant one. Patients who feel pressured to consent will have a worse session, not a better one.
"Is the AI making decisions about my treatment?"
"No. The AI only helps me write the note — it documents what we discussed. All clinical decisions, diagnoses, and treatment plans are mine. The AI doesn't analyze your condition or recommend treatments."
"Will my information be used to train AI models?"
This is the question patients are increasingly asking, and the answer needs to be honest. If your AI scribe vendor uses session data for model training, you need to tell patients that. If they don't (and Scriber doesn't), the answer is straightforward:
"No. Your session data is used only to generate your clinical note. It's not used to train AI models, and it's not shared with any third party for research or product development."
Consent Best Practices
- Get verbal consent at the start of each session. A one-time consent form is legally sufficient in most states, but verbal consent at the start of each session is better practice — it respects the patient's right to change their mind and keeps the relationship transparent.
- Document the consent in the note. A single line — "Patient consented to use of AI scribe for this session" — creates a record that the consent was obtained.
- Know your state's recording laws. Our state-by-state telehealth recording guide covers the consent requirements (one-party vs. two-party consent) that apply to AI scribe recordings. Most states require only one-party consent, but a handful require all parties to consent.
- Have a written consent form for new patients. During intake, include a brief explanation of the AI scribe in the practice's consent forms. This gives patients time to consider it before the first session, and gives you a signed document if a compliance question ever arises.
- Make opting out frictionless. If a patient says no, switch to manual notes without commentary. The patient should never feel that declining the AI scribe inconvenienced the clinician.
When Not to Use an AI Scribe
There are sessions where skipping the scribe is the right call, and naming those proactively builds trust:
- First-session crisis or safety assessment. If a patient is in acute distress, the presence of a recording — even if the patient consents — can change the dynamic. Consider taking manual notes for these sessions.
- Patients who have explicitly requested no recording in the past. Respect prior preferences without re-asking each time.
- Sessions involving sensitive content the patient may not want transcribed. Some patients are comfortable with clinical notes but uncomfortable with a full audio recording of certain topics. Let the patient guide this.
The Bigger Picture
Patients are increasingly familiar with AI tools in their own lives — voice assistants, transcription apps, AI features in consumer products. The clinicians who have the smoothest introductions are the ones who treat the AI scribe as a normal part of their workflow rather than something that needs a lengthy disclaimer. Confidence is contagious; over-explanation creates suspicion where none would have existed.
For more on the compliance side of this conversation, our HIPAA primer for AI scribe evaluation covers what to verify with a vendor before you ever introduce the tool to a patient.
Want to try Scriber with your patients? Start a 7-day trial — the BAA is included at signup, and you can see how the introduction feels in your own sessions.