A Sydney-based AI startup scrapes a few million publicly accessible profile photos and short bios to fine-tune a face-matching model it sells to corporate security teams. Annual revenue is under $1.5 million, so the company has spent the last two years comfortably outside the Privacy Act’s $3 million APP entity threshold and treats privacy compliance as a Series A problem. In April 2026 a Melbourne user discovers her face is being used to demo the product to an enterprise customer, instructs a no-win-no-fee firm, and files in the Federal Court the next week — not under the Privacy Act, but under the new statutory tort of serious invasion of privacy. The company has no privacy notice, no opt-out, and no insurance carve-out for the claim. The maximum damages exposure for non-economic loss alone is $478,550, plus economic loss, plus the cost of an interlocutory injunction restraining further use of the model.
That scenario is no longer hypothetical. The statutory tort created by Schedule 2 of the Privacy and Other Legislation Amendment Act 2024 (Cth) commenced on 10 June 2025 and inserted a brand-new Schedule 2 into the Privacy Act 1988 (Cth). It is the most significant expansion of private-litigation risk in Australian privacy law in three decades, and — critically for early-stage companies — it has no turnover threshold, no APP-entity precondition, and no requirement to lodge an OAIC complaint first. Any individual can sue any company in any superior court, full stop.
What the Tort Actually Covers
The cause of action under Schedule 2 has two limbs. A plaintiff must establish that the defendant either:
- intruded upon their seclusion — physical or technological intrusion into a private space, communications or activities, such as covert recording, location tracking, unauthorised access to accounts or premises; or
- misused information relating to them — collecting, using or disclosing personal information in a way that defeats a reasonable expectation of privacy.
The plaintiff must then satisfy four further elements: a reasonable expectation of privacy in all the circumstances; that the invasion was serious; that the conduct was intentional or reckless (negligence alone is not enough); and that the public interest in privacy outweighs any countervailing public interest such as freedom of expression, the proper administration of government, or public health and safety.
The tort is actionable per se — the plaintiff does not have to prove actual damage. Hurt feelings, anxiety and reputational distress are recoverable. So is economic loss where it can be evidenced.
Damages, Limitation, and Standing
The plaintiff-side economics are what make this tort a live litigation risk rather than a theoretical one.
- Non-economic loss is capped at $478,550 (or the equivalent defamation cap if higher), with exemplary or punitive damages available in exceptional cases inside the same cap.
- Economic loss sits outside the cap. A startup whose conduct destroys an individual’s livelihood, business or property interest faces uncapped financial exposure on that head.
- Injunctive relief is available — including interlocutory injunctions, as confirmed in Kurraba Group Pty Ltd & Anor v Williams [2025] NSWDC 396, the first published decision under the new tort, handed down by Justice Gibson in October 2025.
- The limitation period is short: one year from the date the plaintiff becomes aware of the invasion, or three years from the date of the invasion, whichever is earlier (extended to age 21 for minors).
- Standing is broad. Any individual can sue. The defendant can be a company, an individual director, an employee, or a contractor. There is no minimum-turnover gateway and no requirement that the defendant be an APP entity.
The last point is the one founders most often miss. The Privacy Act itself still exempts most small businesses with annual turnover under $3 million. The new tort does not. A pre-seed startup with two co-founders and no revenue is exposed on identical terms to a listed company.
The First Cases Are Already Running
Two cases have tested the tort within its first six months. The Groth proceeding in the Federal Court (filed by Victorian Liberal deputy leader Sam Groth and his wife against the Herald Sun in August 2025) settled in November 2025 before the journalism exemption was tested. Kurraba Group v Williams (NSWDC, October 2025) is the more instructive of the two for founders: a property-development company and its CEO obtained interlocutory injunctive relief against a neighbour who had published private wedding photographs on a vendetta website in connection with a development-approval dispute. The case demonstrates that the tort is being used not just against media defendants but in commercial-dispute contexts where private information is weaponised — exactly the territory in which startups routinely find themselves.
Several plaintiff-side firms are now openly building practices around the new cause of action. Founders should assume that any future employee dispute, customer complaint, data breach or competitor stoush involving personal information will be assessed against the tort by the other side’s lawyers.
The Five Live Risk Vectors for Startups
The exposure scenarios that come up most often in our advisory work cluster around five fact patterns:
1. AI training data. Scraping personal information — faces, voices, social posts, professional bios — to train or fine-tune models is the highest-risk current vector. The conduct is intentional, the seriousness threshold is increasingly met where models reproduce identifiable individuals, and the reasonable-expectation analysis is moving against scraper defendants internationally. Data licensing agreements for AI startups and clear provenance records are no longer optional.
2. Employee monitoring and offboarding. Covert surveillance, accessing a departed employee’s personal accounts, retaining personal data after termination, and reviewing private communications without a clear policy basis all sit inside the intrusion upon seclusion limb. Sham-contracting fact patterns compound the risk because contractors are individuals with a stronger expectation of privacy than employees.
3. Customer data leaks. A breach is not automatically a serious invasion — but the tort focuses on the conduct that caused the disclosure. A startup that chose to store sensitive personal information in a system it knew was inadequately secured is at recklessness risk even on facts that would once have been treated as negligent.
4. Marketing and re-identification. Re-targeting, behavioural advertising, fingerprinting and the use of “de-identified” data sets that can be re-identified all map onto the misuse of information limb. Startups buying data sets from upstream vendors should ask hard provenance questions.
5. Product features that intrude. Dating-app screenshots, fintech transaction-sharing, healthtech symptom logs, location-based services, voice assistants — anything where the product itself processes information a user reasonably expects to stay private. Design choices made at the MVP stage will be litigated later.
What Founders Should Do Now
Re-paper your privacy notice. Reasonable expectation of privacy is heavily shaped by what users were told. A clear, granular privacy notice — explaining collection, use, retention, sharing and AI training — is the cheapest defence available.
Treat the $3 million threshold as irrelevant. If your business model touches personal information, comply with the Australian Privacy Principles substantively even if the Act does not technically apply. The tort will be measured against community expectations, not against your turnover.
Review your D&O and cyber policies. Many older policies do not respond to statutory-tort claims, particularly where conduct is alleged to be intentional. Check the insured-versus-insured exclusion, the intentional-acts exclusion, and the definition of “claim” — and ask your broker for a written confirmation of cover specifically for Schedule 2 claims.
Lock down AI training data provenance. Maintain a register of data sources, licences, consent positions and de-identification methods. If the data set was scraped, the question is no longer “is it legal?” but “is the conduct reckless given a reasonable expectation of privacy?”
Diary the journalism exemption carefully. Media-adjacent startups (newsletters, podcasts, creator platforms, citizen journalism tools) can rely on the journalism exemption, but only for conduct that is genuinely journalism — not for commercial product features that happen to publish content.
Treat departing employees and disgruntled customers as potential plaintiffs. The one-year limitation runs from awareness, so the window is short — but interlocutory injunctions can land within weeks. Build incident-response playbooks that move at that pace.
The Bottom Line
The statutory tort of serious invasion of privacy is the first piece of Australian privacy law that gives an individual a direct, well-funded path into court against a startup, with a $478,550 floor on non-economic damages and uncapped economic loss on top. The first decisions have confirmed that injunctive relief is real and quick. The exposure attaches to conduct, not size — and the conduct categories that bite hardest (AI training data, employee monitoring, customer data handling, behavioural marketing, intrusive product features) sit squarely in the operating model of most early-stage technology companies. Founders should rebuild their privacy posture on the assumption that the next privacy complaint will arrive as a statement of claim, not an OAIC letter.
Viridian Lawyers advises Australian startup founders, boards and investors on privacy compliance, data-handling risk and the new statutory tort of serious invasion of privacy. If you are building an AI product, handling sensitive customer data, or worried about your existing privacy posture under the Schedule 2 regime, get in touch.
