AI scribes in Australia draw privacy warning for doctors

AI scribes are moving through Australian GP clinics faster than regulators have settled the rules for them, prompting a federal warning to doctors about privacy, consent and clinical safety. The Australian Department of Health and Aged Care has urged clinicians to slow their rollout and check their obligations while the Therapeutic Goods Administration reviews whether some products should face tighter regulation. The Guardian reported that the warning follows a sharp rise in use across general practice.

In comments reported by The Guardian, the department said current AI scribing products “have little oversight” even though they can record consultations, generate notes and feed information into digital health systems. Officials said doctors must make sure patients understand what the software does, where their data goes and who remains responsible when a transcript or summary is wrong.

“Our position is that informed consent requires consumers to understand the benefits and limitations of the technology to which they are consenting.”

Australian Department of Health and Aged Care, via The Guardian

Adoption has moved quickly enough to draw Canberra’s attention. The Guardian cited a Royal Australian College of General Practitioners poll showing 22 per cent of Australian doctors were using AI scribes in August 2024, rising to 40 per cent by November 2025. Vendors have marketed the software as a way to cut paperwork and, in some cases, lift practice revenue by as much as 30 per cent, while some providers say they have processed hundreds of millions of consultations globally.

The regulatory gap is sharp because medical scribes sit inside the appointment itself. Unlike a generic chatbot used after hours, the software helps shape a note that may later be billed against, shared with other clinicians or uploaded into digital health systems. Even where a doctor reviews the draft, the tool has already entered a workflow with privacy, record-keeping and safety consequences.

Oversight catches up

Official guidance points to a stricter operating model than the sales pitch suggests. Victoria’s health department guidance on artificial intelligence in public health services says services must assess privacy, cyber security, procurement and human oversight before deploying AI tools. It treats clinical accountability as a governance issue for health services, not something that can be handed off to a vendor.

The Victorian framework also shows how regulators are likely to classify these products in practice. Rather than treating scribes as harmless back-office software, the guidance places them in the same governance conversation as other clinical AI systems: risk assessment before rollout, clear lines of human review and documented controls around patient information. For hospitals and clinics already trialling scribes, that is a tougher standard than a promise of faster notes.

The same caution runs through the RACGP’s medicolegal analysis, which says GPs need to verify outputs, understand contractual and storage arrangements, and obtain proper patient consent before using AI in consultations. The college frames the technology as an aid, not a decision-maker. If a note is inaccurate, incomplete or stored in a way that breaches privacy obligations, responsibility still sits with the clinician and practice, not the software provider.

Health officials told The Guardian the risk is broader than an awkward summary or a missed symptom. If AI-generated notes are uploaded into national digital health infrastructure or used to support later care decisions, errors and unclear consent trails can follow the patient beyond a single appointment.

“This has implications for patient safety, clinical accountability, and the integrity of data held within national digital health infrastructure.”

Australian Department of Health and Aged Care, via The Guardian

The warning shows Australian AI policy moving from general principles to frontline professional use. In healthcare, the government’s message is that efficiency gains are not enough. Until the regulatory position is settled, doctors will be expected to explain the tools clearly, secure meaningful consent and check every output before it becomes part of a patient’s record.