Ted's Friday Five — WTF Happened in AI Allied Health This Week? Week ending 19 June 2026
Jun 19, 2026[VIDEO EMBED — Add recorded video here before publishing]
[PDF DOWNLOAD LINK — Upload TedsFridayFive_20260619.pdf and link here before publishing]
Ted Jedynak, the AI for Allied Health Guy · Helping Clinic Owners Implement AI Safely & Effectively for Better Clinical Outcomes.
Two regulatory deadlines are now running simultaneously toward the end of 2026, and this week a free AI tool launched that clinically outperforms human doctors — with zero training infrastructure attached. If you're an allied health clinic owner in Australia who's adopted any AI in the last 12 months, this week's signals matter more than most.
Signal 1: Your AI Scribe Might Be an Unregistered Medical Device
The TGA issued a formal classification clarification on 30 January 2026. Most Australian clinic owners still haven't read it — but they will when the December 11 enforcement deadline lands.
Here's the rule: AI scribing tools that transcribe and convert clinical conversations into written notes are not medical devices under Australian law. AI scribing tools that analyse clinical content and generate diagnoses, differential diagnoses, or treatment recommendations that the practitioner didn't explicitly state — those are medical devices. They must be listed on the Australian Register of Therapeutic Goods (ARTG) before being used in a clinical setting.
The TGA has already run its first compliance sweep. This is no longer a theoretical risk.
The non-obvious problem is the feature arms race happening inside AI scribe products right now. Vendors like Heidi, Lyrebird, and others are adding "intelligent suggestions" in software updates — often without making it obvious in their marketing. A tool that sat clearly on the "administrative scribe only" side of the line when you signed up may have crossed into medical device territory since your last software update. The classification follows function, not label.
Why this matters to your clinic: If you're using an AI scribe that includes diagnostic suggestions and it's not on the ARTG, you're exposed — not the vendor. AHPRA's professional obligations rest with the practitioner, not the software company. December 11, 2026 is the hard stop.
Your move: Check your AI scribe's full feature list today. If it includes any diagnostic, differential, or treatment suggestions, ask your vendor for their ARTG registration number. No number — you have a problem that needs fixing before 11 December 2026.
Signal 2: December 10 — The Other Compliance Deadline Most Clinics Don't Know About
One day before the TGA's AI scribe deadline, a different legal obligation kicks in. On 10 December 2026, automated decision-making (ADM) transparency obligations under Australia's Privacy Act 1988 come into force.
The law requires that any healthcare practice using a computer program to make or contribute to decisions that "could reasonably be expected to significantly affect the rights or interests" of a patient must disclose how personal information is used in that process. That language covers AI booking triage, intake screening, appointment prioritisation, risk flagging tools, and potentially any AI-assisted clinical documentation system.
The critical detail that most clinic owners will miss: health service providers are explicitly excluded from the small business exemption. A sole-practitioner podiatry clinic in Adelaide has the same Privacy Act obligations as a major hospital group. There is no size threshold that lets you opt out.
Why this matters to your clinic: Privacy compliance consultants and legal firms will start marketing to allied health clinics about this deadline by August. They'll frame it as a legal problem requiring a legal solution. The smarter position is to understand what compliance actually looks like inside your clinical workflows before the lawyers arrive. The CI Program covers exactly what this looks like in practice.
Your move: Audit every automated system in your clinic that touches patient data. AI booking, automated recalls, intake questionnaires, triage screening — all of it. Then update your privacy policy and consent forms to disclose AI involvement before 10 December 2026.
Signal 3: ChatGPT for Clinicians Is Free, Live, and Outperforms Doctors — Without Any Training Attached
OpenAI's ChatGPT for Clinicians launched this week. Powered by GPT-5.4, free for verified practitioners, it scored 59.0 on the HealthBench Professional clinical benchmark. Human doctors given unlimited time and full internet access scored 43.7. That gap is not a typo.
The enterprise version — ChatGPT for Healthcare — is already deployed at Stanford Medicine Children's Health, Memorial Sloan Kettering, and HCA Healthcare. For Australian allied health practitioners: there's no geographic restriction on the tool. It's accessible right now.
The implementation gap this creates is specific and urgent. OpenAI has removed the cost barrier and the access friction simultaneously. But there's no AHPRA-specific guidance attached to the product. No training on when to trust the output and when not to. No framework for informed consent when an Australian practitioner uses a US-designed tool to inform a clinical decision. No consideration of how AHPRA's professional judgement obligations apply when a system outperforms doctors on benchmarks.
Why this matters to your clinic: Your patients are already using this. Some will arrive having run their symptoms through ChatGPT for Clinicians and formed strong views about their diagnosis. The clinic owners who have a considered, professional framework for handling that conversation will be in a categorically different position to the ones who haven't thought about it yet.
Your move: Test ChatGPT for Clinicians yourself this week. Spend 30 minutes on it. Run a few clinical scenarios. Form your own professional opinion about where it's genuinely useful and where it falls short — before your patients have that opinion ready-made when they walk in.
Signal 4: 81% of Doctors Use AI, 70% Got Zero Training — That's the Entire Market Argument
The Philips Future Health Index 2026 and AMA Physician Survey 2026 both landed this week with the same finding: 81% of doctors now use AI in practice — more than double the 2023 rate. Clinicians are saving an average of 132+ hours per year — more than three full working weeks. Some report seeing eight additional patients per week as a direct consequence of AI-assisted documentation.
But here's the number that matters most for your market: 70% of clinicians report that AI training at their organisation is inadequate, inconsistent, or simply not available.
This is not a theoretical market opportunity. It's a documented, evidence-validated gap between what clinicians have access to and what they know how to use responsibly. Australian allied health private practice tracks 12–18 months behind medical adoption curves. The adoption cliff is not coming — it's here.
Why this matters to your clinic: The clinic owners who build structured AI protocols now, with proper clinical governance, will be ahead of mandatory competency requirements when AHPRA formalises them — which is when, not if.
Your move: Pull the Philips stat — "132 hours saved per year, but 70% of clinicians get zero training" — into your next patient-facing or referrer-facing communication. It's citation-ready from a credible global survey and makes the case for structured implementation better than any marketing copy.
The Bottom Line This Week
Two compliance clocks are ticking toward December 2026, a clinical AI tool just crossed the performance threshold where it outperforms unassisted human doctors, and the data confirming the training gap is now peer-validated and citation-ready. The direction of travel for AI in Australian allied health is unmistakable: adoption is accelerating, regulation is tightening, and the training infrastructure is nowhere near keeping up. The clinic owners who act on the signals now — not in August when the compliance consultants start calling — will be the ones who don't have to scramble.
Ted Jedynak is a podiatrist, educator, and business coach based in Adelaide, Australia. His Clinical Intelligence Program (AI+HI=CI) helps allied health clinic owners implement AI into their practice — practically, safely, and without the hype. Find out more at tedjedynak.com/clinical-intelligence
Sources: TGA SaMD Classification Guidance Jan 2026 (tga.gov.au) · Privacy Act 1988 (Cth) ADM Amendments Dec 2026 · Philips Future Health Index 2026 (philips.com/futurehealthindex) · AMA Physician Survey 2026 (ama-assn.org) · OpenAI ChatGPT for Clinicians/Healthcare launch Jun 2026 (openai.com/health) · HealthBench Professional Benchmark Jun 2026 · AHPRA AI Guidance (ahpra.gov.au)
