Built to the standard AHPRA expects.
Every clinical AI feature in RecovrFlow is gated by an explicit, logged, withdrawable patient consent. Aligned with AHPRA's August 2024 AI guidance and Australian state surveillance laws.
Consent isn't optional. It's the first thing AHPRA looks at.
AHPRA's August 2024 statement set five principles for any registered practitioner using AI in clinical practice: Accountability, Understanding, Transparency, Informed Consent, and Ethical/Legal use. Of those, informed consent is the one that fails most quietly \u2014 a clinician trials an AI scribe, forgets to mention it, and the record has no trail.
Layered on top is Australian state law. NSW, SA, WA, ACT, and Tasmania are all-party consent jurisdictions \u2014 you cannot record a consultation without the patient's explicit agreement. Skipping consent isn't just an AHPRA risk; in those states it's a criminal offence.
RecovrFlow handles both in one flow. No AI runs against a patient's data until you've recorded their consent \u2014 and the record holds up under audit.
Every consent event, the way auditors want to see it.
Method captured
Digital (in-app signature), verbal (clinician confirms patient agreed), or written (paper form scanned in) — every consent event records how it was obtained.
Timestamped and attributed
Every grant or decline is stamped with the exact time and the clinician who recorded it — not the patient's clinician in general, the specific user.
Per-patient, not per-org
Consent is bound to the patient record. Moving a patient between clinicians or organisations does not silently inherit consent — it must be re-confirmed where appropriate.
Clinician notes attached
Free-text notes let the clinician record context — e.g. 'verbal consent obtained, patient declined written form due to vision impairment' — so the consent record stands up under audit.
Immutable audit log
Every consent event writes an AI_CONSENT_RECORDED entry to the audit log. Past events are never overwritten — new events are appended, so the full history is reconstructible.
Withdrawable at any time
Patients can withdraw at any time by telling their clinician, who records the withdrawal in the same flow. The patient's record stays accessible to the clinician — only AI features disable.
Five steps. Then AI either runs, or it doesn't.
- 1
Open the patient
Clinician navigates to a patient record.
- 2
Consent banner appears
If AI consent has not been recorded, a banner sits at the top of the record.
- 3
Open the consent modal
Clinician opens the consent flow, with plain-language explanation visible to share with the patient.
- 4
Grant or decline
Clinician records the patient's decision, including method (digital / verbal / written) and optional notes.
- 5
Gate state updates
AI features unlock (or stay off). Audit log writes an AI_CONSENT_RECORDED event.
Decline means decline. The clinician keeps full manual access to the record \u2014 only AI features switch off. There's no nag, no override, no hidden re-prompting.
How RecovrFlow maps to AHPRA's principles.
| AHPRA principle | What it means | How RecovrFlow meets it |
|---|---|---|
| Informed Consent | AHPRA expects patients to be told AI will be used and to actively agree before it is. | Consent gate fires before any AI feature runs. Decline keeps AI fully off for that patient — the clinician still uses the platform manually. |
| Transparency | Patients should be able to understand what AI is doing with their information. | The consent modal plain-language explains: what AI does, what data is processed, where it is stored, and what 'withdraw' means. |
| Accountability | The treating practitioner remains accountable for clinical decisions, including those informed by AI. | Consent is recorded against a named clinician user. Clinicians review every AI output before saving — nothing publishes automatically. |
| Records | Use of AI in a clinical interaction should be documented. | Consent event, method, time, clinician, and any notes are written to the audit log. Reconstructible if AHPRA, the NDIA, or a court ever asks. |
Source: AHPRA, “Meeting your professional obligations when using Artificial Intelligence in healthcare,” 22 August 2024.
Where this fits with the rest of compliance.
AHPRA consent is one of four pillars. The others are the data layer.
Row-Level Security
Access is restricted to authorised patient records, enforced at the database layer.
See securityNo AI Training on Your Data
Your patient data is not used to train AI models, and temporary processing data is not kept.
See securityAustralian Sovereignty
Data lives exclusively on Australian servers. Privacy Act 1988 + APPs.
See securityAHPRA-Aligned Consent
Explicit, logged, withdrawable patient consent before any AI runs.
You're hereFrequently asked questions.
Is patient consent legally required to use an AI scribe in Australia?+
AHPRA's August 2024 guidance on AI in healthcare states that informed consent is one of five principles practitioners must meet when using AI. Separately, state Surveillance Devices Acts govern audio recording — NSW, SA, WA, ACT, and Tasmania are all-party consent jurisdictions, meaning the patient must agree to being recorded before any audio is captured. RecovrFlow's consent gate covers both bases in one flow.
What if a patient declines AI consent? Can the clinician still see their record?+
Yes. Declining consent does not lock a clinician out of the patient's record — it only disables AI features for that patient. The clinician can still write notes manually, view history, and run the rest of the platform. AI just stays off until the patient changes their mind.
Can a patient withdraw their consent later?+
Yes, at any time. The patient tells their clinician, the clinician records the withdrawal in the same consent flow, and AI features disable immediately for that patient. The withdrawal is logged as a new audit event — the prior consent is not deleted, just superseded.
Is the audit log accessible to me as a clinician?+
Yes — each patient record surfaces the current consent state and the latest event. For deeper review (e.g. AHPRA complaint response, NDIA audit), the full audit history can be exported by the practice owner.
Why is RecovrFlow 'AHPRA-Aligned' rather than 'AHPRA-Certified'?+
AHPRA does not certify software. No vendor can credibly claim AHPRA certification, because the framework does not exist. 'AHPRA-Aligned' means our consent, transparency, accountability, and records design maps directly to the five principles in AHPRA's August 2024 AI guidance. Read the guidance and judge for yourself — we link to it below.
Documentation that holds up. Consent that holds up.
RecovrFlow is the AI clinical documentation platform built for Australian allied health \u2014 including the consent gate that AHPRA expects and state surveillance laws require.
RecovrFlow is not a substitute for clinical judgment. Always review AI-generated outputs before saving to the patient record.