AI scribes are recording your doctor visits — here’s what to know about your privacy

If you’ve visited a doctor’s office lately and noticed the physician typing less and talking more to a screen, you might have encountered an AI scribe. These tools listen to appointments, transcribe conversations, and generate clinical notes automatically. Adoption has surged in the past year — doctors report it saves them hours of paperwork. But a recent warning from the Australian government is raising alarms about what happens to that recorded data.

What happened

In July 2026, the Australian government issued a public warning about the privacy risks associated with AI scribes used in healthcare. The notice, reported by The Guardian, highlighted that many of these services transmit audio or transcripts to third-party cloud servers for processing. In some cases, patients are not explicitly told that their consultation is being recorded, and consent is obtained only in vague terms buried in general practice forms.

The warning was prompted by growing use of such tools across Australian clinics and hospitals. The government pointed out that data may be stored overseas, used to train AI models, or retained longer than necessary. While no specific breach was cited, the scale of the rollout prompted regulators to act before problems become widespread.

Why it matters

AI scribes are not just automated note-takers. They often rely on large language models that need to process the full conversation to generate a summary. This means everything you say during a consultation — symptoms, medications, personal history, even offhand remarks — could leave your doctor’s local system.

Key privacy concerns include:

  • Consent is not always informed. You may have signed a general privacy consent form at your first visit. That form likely does not mention AI transcription. In many cases, doctors themselves may not fully understand how the tool handles data.
  • Data may be stored on third-party servers. Depending on the vendor, recordings or transcripts could be sent to cloud infrastructure in another country, subject to different data protection laws.
  • Potential for secondary use. Some AI scribe providers have terms that allow them to use anonymised data for model training. “Anonymised” is not always irreversible, especially with rich medical data.
  • Retention periods vary. Automatic deletion policies differ. A transcript might be kept for months or years, creating a larger target for breaches.

The Australian warning is important because it is one of the first government-level statements on this specific technology. It signals that regulators are watching, but it also highlights that protections are not yet in place for most patients.

What readers can do

You don’t have to accept AI scribe use without knowing the details. Here are practical steps:

  1. Ask before the appointment. When booking, ask: “Do you use any AI tools to record or transcribe consultations?” Many clinics will say yes — and that gives you the chance to discuss it beforehand.

  2. Request explicit information. If they do use one, ask:
    – Where is the data processed and stored?
    – How long is it kept?
    – Is it used for training?
    – Can you opt out without affecting care?

  3. Opt out if you’re uncomfortable. Under Australian privacy law (and similar laws in many countries), you generally have the right to refuse certain data processing. A clinic should offer an alternative — for example, manual note-taking.

  4. Read the privacy policy. The clinic’s privacy policy should mention any third-party AI services. If it’s vague, consider switching providers.

  5. Consider requesting a copy of your notes. If an AI scribe was used, you can ask for the transcript or summary. That gives you a chance to see what was recorded and whether it matches your recollection.

These steps may feel awkward, but they are reasonable. Doctors are still learning about the privacy trade-offs themselves.

Sources

  • “Doctors’ soaring use of AI scribes prompts Australian government warning over privacy,” The Guardian, 5 July 2026. Link

(Note: The link is from Google News and may require a referral. For the original article, search “AI scribes Australian government warning privacy” on The Guardian’s website.)