The Privacy Act changes affecting Australian clinics are not all new, and they are not all in force. A patient can already sue your practice directly, a disclosure rule starts on 10 December 2026, and in June the Privacy Commissioner found two health providers broke the law through website tracking. This guide sorts law from proposal, then shows how to check your website, ads and AI tools.
The short answer
You are covered by the Privacy Act whatever your turnover. One fixed statutory date applies to most clinics (10 December 2026), and it is conditional. The bigger practical change in 2026 is existing law applied to health websites and advertising, and a draft second tranche of reform is not law.
- You are covered. Private health service providers are inside the Privacy Act. The small-business exemption debate does not apply to you.
- Patients can already sue. Since 10 June 2025 a patient can bring a claim for a serious invasion of privacy without going to the Office of the Australian Information Commissioner (OAIC) first.
- 10 December 2026 is the one fixed date. If software your clinic uses makes, or substantially supports, decisions that could significantly affect patients, using their personal information, your privacy policy must say so. Some clinics may find nothing to disclose, but should record why.
- The OAIC applied existing law to health websites. Privacy Commissioner Carly Kind found in June 2026 that two health providers breached the Act through tracking pixels and retargeting. A clinic stays responsible for tools it commissions.
- The second tranche is a draft, not law, and it does not remove the small-business exemption.
- Do three things first. List every tag, form and AI tool you use. Pause patient-list uploads and advertising tags on pages that reveal health information until someone qualified has reviewed them. Assess each tool against the 10 December test and write down the answer.
How to read this guide. Where we describe how the law applies, that is our reading as a marketing agency, not legal advice, and some points are unsettled. We flag the unsettled ones and name what to ask a privacy lawyer.
What is law, what is due on 10 December, and what is only proposed
Most advice mixes up four different things: laws already in force, one commencement date, a draft Bill, and a few claims that are wrong. This table separates them as at November 2026.
| Status | Item | What it means for a clinic |
|---|---|---|
| In force | Privacy and Other Legislation Amendment Act 2024. Most of it commenced on 11 December 2024, bringing tiered penalties, infringement notices and compliance notices. | The OAIC can act on lower-level failures, including a privacy policy that does not meet the rules, without going to court. |
| In force | Statutory tort for serious invasions of privacy, from 10 June 2025. | A patient can sue a clinic directly. |
| In force | $5.8 million civil penalty against Australian Clinical Labs, ordered by the Federal Court on 8 October 2025. | The first civil penalty under the Act. Penalties reach health providers. |
| In force (existing law applied) | OAIC determinations against Monash IVF and Medmate, 11 June 2026. | The most practical 2026 development for clinic websites and advertising. Covered below. |
| Due 10 December 2026 | New Australian Privacy Principles (APPs) 1.7 to 1.9: privacy policies must describe certain decisions made or supported by computer programs. The OAIC issued final guidance on 30 September 2026. | The only fixed statutory deadline for most clinics, and it is conditional. |
| Proposed, not law | Privacy Amendment (Personal Data Protection) Bill 2026. Exposure draft released 31 August 2026, consultation closed 18 September, and the government says it intends to introduce legislation before the end of 2026. | Plan for its direction. Do not tell staff or patients it applies. |
| Not in the draft | Removing the $3 million small-business exemption, removing the employee-records exemption, a direct right of action for privacy breaches, mandatory privacy impact assessments. | Several commercial articles claim otherwise. |
| Wrongly reported | “The small-business exemption ends on 10 December.” “Clinics will get a right to erasure.” The AML/CTF changes of 1 July 2026 as a privacy change for clinics. | None is correct. 10 December is the automated-decision disclosure date. The draft’s erasure right covers only large digital platforms (for example, social media services with $500 million revenue or 2.5 million users), not clinics. The AML/CTF changes affect accountants, lawyers and agents. |
Has the small-business exemption been removed?
No. The exemption for businesses turning over $3 million or less remains in force and the draft Bill does not remove it. It never covered private health service providers, so it does not change your obligations.
Has the second tranche of reform passed?
No. It exists only as an exposure draft. The government has said it intends to introduce legislation before the end of 2026, and the length of any transition period is unknown.
Does the Privacy Act apply to your clinic?
Yes, if you provide a health service and hold health information, and size does not change that. The OAIC’s guide to health privacy lists GPs and specialists, private hospitals, allied health professionals, pharmacists, gyms and weight-loss clinics, fertility clinics, and health services delivered online, including telehealth.
The $3 million threshold is the general small-business exemption, and private health service providers were carved back into the Act. A dental practice turning over $900,000 (a hypothetical) is covered just as a large group is. Public hospitals sit under state and territory regimes, so check which applies if you work across both.
- Do you provide a health service? If yes, continue.
- Do you hold information about patients’ health or the services they receive? If yes, it is health information and, under the Act, sensitive information. That attracts stricter rules on collection, use and disclosure.
- Does turnover change the answer? No.
What changes on 10 December: the automated decision disclosure
From 10 December 2026, if your clinic arranges for a computer program to make, or do something substantially and directly related to making, a decision that could significantly affect a patient, using their personal information, your privacy policy must describe the kinds of information used and the kinds of decisions involved. If none of your tools meet that test, you have nothing to add under this rule, but you should record why.
The test has three parts, and all three must be met, as set out in the OAIC’s APP 1.7 to 1.9 fact sheet:
- A program makes a decision, or does a thing substantially and directly related to making one.
- The decision could reasonably be expected to significantly affect the individual’s rights or interests.
- Personal information is used in the program’s operation.
This is a transparency rule about your privacy policy. It does not ban any tool, and by itself it does not require patient consent.
Applying the test to real clinic tools
Advisers disagree on the borderline cases. Avant’s lawyers treat AI scribes, triage, recall systems, billing tools that select item numbers and diagnostic support as potentially in scope. Other advisers treat note-only scribes as a grey area. The table below is a cautious reading for your privacy lawyer to confirm.
| Tool | Reading | Why |
|---|---|---|
| AI triage that sorts patients by urgency | Likely in scope | It directly shapes who is seen and when |
| Diagnostic or risk-prediction software | Likely in scope if it substantially supports a clinical decision | The decision significantly affects the patient |
| Recall logic that decides which patients are contacted | Unclear | Depends on whether selection affects access to care |
| Billing software that auto-selects item numbers | Unclear | Affects fees and claims, but a clinician usually confirms |
| Online booking rules that refuse or redirect patients | Unclear | Depends on how much the rule decides without a person |
| AI phone line that books and routes callers | Unclear | Depends on whether it decides appointment type or eligibility |
| AI scribe with clinician review | Unclear, often treated as lower risk | It drafts notes rather than decisions, though disclosure costs little |
| Advertising audience selection | Not obviously in scope; assess it | It is mainly a separate privacy question, covered below |
What to record, and what to write
For each tool, write a short note: what it does, whether a person reviews its output, what personal information it uses, which of the three parts are met, your conclusion (in scope, unclear, or not in scope), and who decided and when. Where the answer is “unclear”, disclosing is usually the lower-risk choice.
Where a tool is in scope, the disclosure follows a simple structure, to be settled with your lawyer:
We use [type of software] to [kind of decision or step it performs]. It uses [kinds of personal information, such as appointment history or clinical notes]. [A clinician reviews its output before it affects your care. / No clinician reviews this step.]
The OAIC’s fact sheet explains the level of detail it expects. The OAIC can now issue infringement notices for lower-level failures, including a privacy policy that does not meet the rules. Be cautious about dollar figures on vendor blogs, and check any number against the legislation before you repeat it.
Tracking pixels on clinic websites: what the OAIC found
A tracking pixel is a snippet of code from an advertising or analytics platform, such as the Meta Pixel, Google Ads or TikTok, that sends information about website visitors back to that platform. Pixels are not banned. But in June 2026 the Privacy Commissioner found that two health providers, Medmate and Monash IVF, breached the Privacy Act by using them on health websites without consent.
The OAIC’s media release on the determinations announces the outcome. The details below are drawn from the determinations as reported in legal commentary. Both providers were ordered to stop collecting the data within 60 days, destroy it where the law permits, tell the OAIC within 90 days that they had complied, and obtain consent and give notice before using pixels again.
What the Commissioner found
- Browsing a health website can be health information. Page-level activity on a health provider’s site can allow an opinion to be formed about a visitor’s health. That makes it sensitive information.
- Having no names did not help. The absence of direct identifiers did not stop the data being personal information.
- Retargeting counted as direct marketing. Showing ads to individuals based on their behaviour on the website engaged the direct-marketing rule. For sensitive information, that requires consent.
- Three principles were breached: APP 3.3 (collecting sensitive information), APP 5.1 (notifying people about collection) and APP 7.1 (direct marketing).
- A policy that mentioned marketing did not satisfy notification. A privacy policy alone was not enough.
- Neither provider had completed a privacy impact assessment before deploying tracking. A privacy impact assessment is a structured review of how a project handles personal information, done before launch.
- The scale was not small. Monash IVF ran seven different pixels at various times (Meta, Google Ads, Google Analytics 4, Matomo, Jet Interactive, Hotjar and Pinterest) between 2012 and December 2024. Medmate’s TikTok pixel URLs contained the conditions and medications visitors were looking at. Monash IVF also uploaded Custom Audience lists, which are customer lists loaded into an ad platform so the platform can find and target the same people, containing names and contact details to Meta, and could not tell the OAIC where the data came from.
- It began with a scan. The OAIC inspected 50 health-provider websites in late 2024 and contacted 12 providers. Several did not know pixels were active, and one found 50 active pixels it did not need. The OAIC’s blog post Your life, pixelated describes the exercise.
What this means for your clinic
- The finding is not limited to fertility or telehealth. Any clinic website that sends page-level data to advertising platforms may be collecting sensitive information, depending on what the pages reveal.
- The risk sits in specific pages and uses. A pixel on a home page is different from a pixel on a condition page, a booking page, or a confirmation page that shows the service booked. If your paid search campaigns build remarketing lists from visits to those pages, they sit in the same territory.
- Moving the same data differently does not fix it. Server-side tagging or a conversion API changes how data travels. It does not change what is collected, who receives it or why. This is expert interpretation, not a finding.
- Hashing is not anonymisation. Platforms match hashed emails and phone numbers against their own accounts so advertisers can reach existing customers. Information that can be matched back to a person is not de-identified, and only genuinely de-identified information can fall outside the Act.
- A cookie banner is a starting point, not a fix. Some commentary on the determinations says a banner alone was not enough. A banner helps only if tags stay off until a visitor makes an informed, specific choice. One that appears after tags have fired, or bundles every purpose into one “accept” button, is hard to defend.
Where patient information leaves a clinic through marketing and enquiry tools

Use this as your map. The Australian Privacy Principles (APPs) named in the middle column are the ones most often engaged. APP 8 covers disclosure overseas and APP 11 covers security. If a vendor holds patient data offshore, the clinic stays accountable for how that vendor handles it.
| Step | What moves, and which APPs engage | Typical failure |
|---|---|---|
| Ad click and landing page | Page address, device and ad identifiers (APPs 3, 5, 7) | Tags fire before any consent choice |
| Service and condition pages | Page names that reveal a health concern (APPs 3.3, 7) | Retargeting audiences built from these pages |
| Booking widget | Booking details, sometimes passed to the platform’s own tags (APPs 3, 5, 6, 8) | The clinic cannot see the platform’s tags |
| Intake and contact forms | Names, contact details, symptoms typed in free text (APPs 3, 5, 6, 11) | Pre-ticked marketing consent; free text sent to analytics or session replay |
| Practice software or CRM | Patient records and contact lists (APPs 6, 11, 12, 13) | Broad staff access; exports to other tools |
| Phone and call tracking | Caller numbers, recordings, transcripts (APPs 3, 5, 8 and state recording laws) | No notice; offshore storage |
| Recall and review messages | Patient contact lists and visit timing (APPs 6, 7 and the Spam Act) | Lists uploaded to vendors without a documented basis |
| AI tools | Audio, notes, prompts and outputs (APPs 3, 5, 6, 8, 11 and the 10 December test) | Staff use personal accounts; data stored overseas |
| Agency offboarding | Admin logins, tags and audiences (APP 11) | Live tags and audiences left behind |
Who is responsible: you, your agency or your software vendor
You remain responsible. On the OAIC’s reasoning as reported, Monash IVF and Medmate held the information because they commissioned the pixels, decided what they captured, and used the data for retargeting, even though it travelled to Meta, Google and TikTok.
Medmate also reviewed its contracts with external marketing agencies during the investigation. A contract can move the cost of a failure to someone else. It cannot move your duties as the regulated entity.
Ask five questions about any tool:
- Who chose it?
- Who decided what it collects?
- Who benefits from the data?
- Who holds the contract?
- Who can switch it off?
If the clinic is the only party that could answer “we do” to all five, responsibility sits with the clinic. If you cannot answer the fifth because you do not have the logins, fix that this week. Careers pages and enquiry forms collect personal details and often carry the same tags, so a careers page belongs in your inventory too.
The draft Bill proposes a controller and processor framework that could change how responsibility is shared between a clinic and its service providers. It is a draft, so it changes nothing today.
Ten questions to put to any agency, web developer or software vendor
- List every tag, pixel and script running on our websites and landing pages.
- What data does each one send, and to whom?
- Does any of it run on condition, treatment, booking or confirmation pages?
- Do you upload any patient or enquiry lists to any platform? If so, on what legal basis?
- How do you capture consent, and how is the choice honoured before tags fire?
- Where is our data stored and processed, including by your subcontractors?
- Is any of it used to train models or improve products for other customers?
- What happens to our tags, audiences and data when we leave?
- How quickly will you tell us about a breach, and what will you tell us?
- What did you change after the OAIC’s June 2026 determinations?
We run advertising for clinics, including Meta campaigns, so these questions apply to us too. Ask us the same ten.
See your website the way a regulator would
The OAIC found these problems by scanning health websites from the outside. You can do a first pass the same way. If nobody on your team is comfortable with a browser’s developer tools, ask your web developer or agency to do it while you watch, and do not accept a written summary alone.
- Open the site in a private window with the browser’s network panel running. Visit the home page, a service page, a condition or treatment page, the booking page and the booking confirmation page. Note every third-party domain contacted, including advertising, analytics and heat-mapping services. A tag manager such as Google Tag Manager is the control panel that loads most of these, so ask who has access to it.
- Look at what travels. In each request, check the page address, the page title and any form data. Look for condition or medication names.
- Check timing. Do requests go to advertising platforms before a visitor has made any choice?
- Compare with your privacy policy. Does it name or accurately describe every tool you found, and where the data goes?
- Check forms. Look for pre-ticked marketing consent, free-text boxes, and where submissions are sent: your CRM, an email platform, an ad audience.
- Check calls and chat. Is there call tracking or recording, a notice at the start of calls, a chat widget that stores transcripts, or an AI phone line?
- Check lists. Where are your recall and review contact lists held, and who uploads them?
- Check access. Who holds admin logins for your tag manager, ad accounts and audiences? Are you one of them?
- Write it all down with the date. That record is part of your evidence of reasonable steps.
A scoring heuristic for each tool
This is Pracxcel’s analytical model, not a legal test. Score each tool from 0 to 2 on five dimensions and add the results.
| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Sensitivity of the page or field | Generic page | Service category | Condition, treatment, booking confirmation or free-text field |
| Identifiability | Aggregate only | Cookie or device identifier | Email, phone, name or hashed match |
| Recipient | Your own system | Australian processor under contract | Third-party platform with its own uses, or overseas |
| Purpose | Clinical or operational | Analytics | Advertising, retargeting or lookalike audiences |
| Consent state | Valid, recorded consent | Notice only | None, or banner only |
- 0 to 3: document it and keep it.
- 4 to 6: fix it by gating it behind consent, stripping page names and parameters, and tightening the contract.
- 7 to 10: stop until it is redesigned and reviewed.
- Hard stop: a 2 for sensitivity, recipient and purpose together means stop, whatever the total.
Sort your pages into three tiers
| Tier | Pages | Rule of thumb |
|---|---|---|
| 1 | Home, about, contact | Ordinary analytics are generally lower risk |
| 2 | Service categories | Review each tag and its purpose |
| 3 | Condition, treatment and booking-confirmation pages, and any page whose address or title names a health concern | No third-party advertising tags without valid consent and legal review |
If you are rebuilding forms, consent and your policy page, healthcare web design is the moment to build consent in from the start rather than bolt it on.
Keeping your marketing measurable
Removing tags from sensitive pages reduces what advertising platforms can report. Judge campaigns against numbers you hold yourself, and keep the tools that feed platforms separate from the ones that stay inside your systems.
| Bucket | Examples | Privacy position |
|---|---|---|
| Platform-measured | Aggregate traffic on Tier 1 and 2 pages where tags remain | Lower risk, subject to review |
| First-party counted | Calls, form submissions, new patients in your practice software, a “how did you hear about us” question at booking, enquiries from local search | Stays inside your systems |
| Do not send back to platforms | Uploads of bookings or patient emails as “offline conversions” | Same disclosure problem as a pixel; get advice first |
Can patients sue your clinic for a privacy breach?
Yes. Since 10 June 2025, a person can sue for a serious invasion of privacy without going to the OAIC first. To succeed, a patient must show:
- They had a reasonable expectation of privacy. Health information sits at the top of that scale.
- The invasion was serious, not merely technical or trivial.
- The conduct was intentional or reckless.
The tort covers intrusion upon seclusion and misuse of information. Defences include that the conduct was required or authorised by law, the patient consented, or the practice reasonably believed it was necessary to prevent a serious threat to life, health or safety. A court can award damages for financial loss and for distress, and can grant injunctions. The Act caps damages for non-economic loss.
The clearest risk is conduct, not software: a staff member opening a patient’s record out of curiosity is intentional access with no clinical purpose. An honest mistake is less likely to meet the test than ignoring a problem you know about. That second point is an interpretation your lawyer should test, and the courts have not yet tested the tort in this setting.
How the exposure builds
| Step | Who starts it | What can follow |
|---|---|---|
| Complaint to the clinic or the OAIC | A patient | Investigation and conciliation |
| Determination | The Privacy Commissioner | Orders such as stopping a practice, destroying data and notifying the OAIC |
| Infringement or compliance notice | The OAIC | Applies to lower-level failures, including a deficient privacy policy |
| Civil penalty | The OAIC applies to the Federal Court | The $5.8 million Australian Clinical Labs penalty is the first example |
| Patient claim | The patient, directly | Damages and injunctions |
One event can trigger several steps at once. A breach that exposes patient records can lead to a notifiable data breach report, an OAIC investigation and individual claims.
The basics every clinic needs, and what your privacy policy should say
The Act now says “reasonable steps” to protect personal information include technical and organisational measures. A policy on file is not enough on its own: the controls need to exist and you need to be able to show them.
| Control | Why it matters | Minimum |
|---|---|---|
| Role-based access | Front-desk staff do not need clinical notes | Access matches role; review it when staff change |
| Audit logging | You cannot detect snooping you cannot see | Turn it on in your clinical software and review it |
| Multi-factor authentication | Health data is a frequent breach target | On practice software, email and remote access |
| Named privacy owner | Someone has to be accountable | A named person, not necessarily full-time |
| Incident response plan | Speed matters after a breach | Written steps from detection to notification |
| Vendor due diligence | A vendor’s breach is your problem | Ask the ten questions before you sign |
| Staff training with records | It shows reasonable steps | Annual session, attendance recorded |
| Retention and destruction | Holding less means exposing less | A schedule that follows your state’s records rules |
If you suspect an eligible data breach, you must assess it promptly and within 30 days, and notify the OAIC and affected people if it is likely to cause serious harm.
What valid consent looks like
Clinical consent and marketing consent are different things. A patient who agrees to treatment has not agreed to be retargeted with ads. Valid consent is informed, voluntary, current and specific, which is why pre-ticked boxes, bundled “accept all” buttons and open-ended future uses are hard to defend. Record the choice, and make it as easy to withdraw as it was to give.
What your privacy policy should cover
- The kinds of personal information you collect and hold, and why
- How you collect it, including through your website, forms, calls and tracking tools
- Who you disclose it to, including any overseas recipients
- How patients can access and correct their records
- How to make a complaint and how you handle it
- Whether people can deal with you anonymously or under a pseudonym
- A statement on automated decisions, if the 10 December test is met
The RACGP publishes a privacy policy template for general practices that is a sensible starting point for any health practice.
Contracts and insurance
Ask your professional indemnity insurer whether AI-assisted care and the tools you use are covered. Update practitioner and contractor agreements so they prohibit unapproved tools with patient data. Check that vendor contracts say where data is stored, whether it is used for training, and who is liable for a breach.
AI scribes, AI phone lines and staff using AI
Three separate questions get blurred together in most advice on AI, and they have different answers.
| Question | What it covers | Where to look |
|---|---|---|
| Consent, recording and where the data goes | Informed patient consent, state recording laws, overseas processing (APP 8), vendor contracts | OAIC, RACGP, AHPRA and state law |
| The 10 December disclosure | Whether the tool meets the three-part test | The test above |
| Clinical or device status | Whether the tool is a regulated medical device | Therapeutic Goods Administration (TGA) guidance |
- Recording a conversation without consent is an offence in some Australian jurisdictions. Check the law where you practise. Victoria’s advisory on ambient AI scribes, for example, points to its Surveillance Devices Act alongside privacy law.
- Consent sits with the practice and the practitioner. The Australian Health Practitioner Regulation Agency (AHPRA) expects informed consent for AI scribes, ideally recorded in the health record.
- Check where the data goes. The RACGP’s AI scribes fact sheet tells practices to confirm whether a scribe’s data is processed or stored outside Australia.
- Scribes that only transcribe are generally not medical devices. Tools that analyse content or recommend treatment may be, and then need to be on the Australian Register of Therapeutic Goods.
- Staff using consumer chatbots with patient information is a breach waiting to happen. The OAIC advises against entering personal information, especially sensitive information, into publicly available AI tools.
- Patients who decline. Advocacy groups report clinics refusing patients who will not consent to a scribe. Whether that is lawful has not been tested. Offer a non-AI option and get advice before making care conditional on consent.
For each AI tool, keep a one-page review: what it does, what information it receives, where it is processed and stored, whether data is used for training, how consent is recorded, how it is deleted, and whether it meets the 10 December test.
A three-rule staff policy
- Approved tools only. Name the tools staff may use with patient information, on practice accounts.
- Banned tools. Personal accounts and consumer chatbots are off limits for anything about a patient.
- Raise it without fear. Tell staff who to talk to when they are unsure or have made a mistake.
Clinical use of AI is outside what a marketing agency should advise on. Use the RACGP, AHPRA and your indemnity insurer for that.
Review requests and recall messages
Review-request and recall lists are patient contact data used for a purpose beyond treatment. A reminder about a patient’s own appointment is a care communication. Messages promoting extra services, or sent to lists built for advertising, are closer to direct marketing, and the line is not always clear. Treat promotional recall as marketing: get separate consent, include an easy opt-out, and check the Spam Act.
Tools that automate review requests, including our own review automation, work from patient contact lists. That makes the ten questions as relevant here as anywhere else: where the list is held, who can access it, what the vendor does with it, and what happens when you leave.
Other rules that overlap
| Rule | What to check |
|---|---|
| AHPRA advertising rules (section 133 of the National Law): testimonials, outcome claims and titles | Our AHPRA checklist, and our guide to AHPRA-safe workplace culture content |
| Spam Act 2003: consent, identification and opt-out for recall and review messages | Current guidance from the Australian Communications and Media Authority |
| State and territory health records laws: retention, access and destruction | Your state’s legislation or your insurer’s guidance |
| Surveillance devices laws: recording consultations and calls | The law in each state where you practise |
| My Health Records Act and Rules 2026: security and access policies for connected practices | RACGP guidance |
Privacy law and advertising law are separate. Passing one does not mean passing the other.
Where exposure tends to concentrate
This is our reading, based on how different practices typically use marketing and technology. Start with the pattern closest to yours, and see how services differ across practice types on our specialty marketing pages.
- Mental health practices: consent for AI notes, and any advertising tag on pages that describe conditions. The same question arises for psychology and psychiatry marketing.
- Cosmetic and dermatology clinics: retargeting after treatment pages, before-and-after images and AHPRA advertising rules. See cosmetic medicine marketing.
- Fertility and women’s health: the Monash IVF pattern, with reproductive-health ads built from site behaviour. See gynaecology and obstetrics marketing.
- GPs: AI scribes, My Health Record and recall messaging.
- Dental and allied health: booking widgets, free-text intake forms and review requests.
Your plan for the three weeks before 10 December
Start with what an outsider can see, then the statutory deadline, then internal governance.
| When | Do this | Why this order |
|---|---|---|
| Straight away | Inventory every tag, form, vendor and AI tool. Get admin access to your tag manager and ad accounts. Record what you find, with dates. Freeze new audience uploads. | You cannot fix what you cannot list |
| Next | Remove or gate tags on Tier 3 pages. Pause patient-list uploads. Switch off session replay on forms. Take advice before restarting retargeting. | These match what the OAIC acted on, and they are visible from outside |
| Then | Check your privacy policy against what you found. Complete the 10 December assessment for each tool and write down the answers. | The one fixed statutory date |
| Then | Separate marketing consent from clinical consent. Record AI scribe consent each time. | Clinical consent does not carry marketing consent |
| Then | Put the ten questions to vendors and agencies, and fix contract gaps. Decide how you will judge marketing without the removed tags, starting from what a new patient is worth to you. The ROI calculator is one way to work that out. | Allocates cost and keeps spend accountable |
| Last | Train staff and file your evidence. Publish the updated policy, ideally a week early. Diarise a review after 10 December and when the Bill is introduced. | Shows reasonable steps |
Keep an evidence file: the dated tag inventory, contracts, consent logs, policy version history, your 10 December assessments, AI tool reviews, a privacy impact assessment for any tracking you keep, and training attendance.
What can wait: detailed second-tranche preparation, and for most clinics the Children’s Online Privacy Code due on the same date.
If the draft Bill passes: design for it now
Do not wait for it, because existing law already applies. But design for its direction. The exposure draft proposes about 40 reforms. Those most relevant to a clinic, as described in law firm summaries, include:
| Proposal | What it could mean for a clinic |
|---|---|
| A “fair and reasonable” test, replacing APPs 3, 4 and 6 | Consent alone would not justify handling that is unfair or unreasonable |
| Consent that must be voluntary, informed, current, specific and unambiguous | Pre-ticked boxes, bundled consents and open-ended future uses become harder to defend |
| Personal information that “relates to” an individual, including information that singles a person out, for example through online tracking | A wider reach for identifiers and tracking data, which matters for pixels |
| Stronger security and destruction duties | Delete what you no longer need |
| A controller and processor framework | Clearer lines between clinics and service providers |
| Faster breach reporting (one firm’s summary reports a 72-hour deadline) | Possibly a much tighter clock than today’s 30-day assessment period |
The draft does not remove the small-business exemption, create a direct right of action, or make privacy impact assessments mandatory. Its erasure right applies only to large digital platforms. In practice: collect less, delete sooner, ask for specific consent, and keep marketing consent separate.
Where Pracxcel fits
We are a healthcare marketing agency, not a law firm, and this guide is not legal advice. We run advertising for clinics, which is the activity this guide examines, so weigh what you have read with that in mind and put the ten questions to us as well.
If you would like to talk through how your website, ads and enquiry handling move patient information, get in touch. For the legal calls, speak to a privacy lawyer.







