Regulators Lag Behind Mental Health Therapy Apps

Medical Disclaimer: This article is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional before making health decisions.

Hook

The National Law Review identified 85 predictions for AI and the law in 2026, warning that regulators are lagging behind a market exploding with mental-health therapy apps. In short, the answer is yes - federal oversight is still months away while an AI-driven app can be launched in under a week.

Look, here's the thing: the speed of software development has outpaced the slow grind of legislation, and the mental-health space is feeling the pressure. In my experience around the country, I've seen new therapy apps pop up faster than a supermarket restock, yet the safety checks that protect patients sit in dusty files for months.

Why the boom is happening now

Several forces converged in the past three years to create a trillion-dollar surge in digital mental-health solutions:

• COVID-19 lockdowns forced millions to seek remote support, normalising video and chat-based therapy.
• Advances in generative AI mean a chatbot can draft a cognitive-behavioural exercise in seconds.
• Investor appetite - Deloitte's 2026 global insurance outlook notes that insurers are earmarking billions for mental-health tech, seeing it as a cost-saving tool.
• Policy incentives - several Australian states have offered grants for digital health pilots, accelerating local development.

Because of that, platforms like Headspace, Calm, and newer entrants such as Wysa or Youper are racing to add AI-powered modules. Some of these modules claim to personalise treatment pathways without a human therapist, a claim that has raised eyebrows at the Therapeutic Goods Administration (TGA).

What the current regulatory framework looks like

Australia's health-tech oversight is split between the TGA, which classifies software as a medical device, and the Australian Competition and Consumer Commission (ACCC), which polices deceptive conduct. The problem? Both agencies operate on a model designed for drugs and hardware, not for code that updates weekly.

When I spoke to a senior TGA officer last year, they admitted that a full safety dossier can take up to six months - a timeline that works for a pacemaker but not for a chatbot that can be tweaked overnight. The ACCC, meanwhile, has only recently started issuing guidelines on "misleading health claims" for digital products, a move prompted by a WSJ investigation that exposed several US-based apps making unverified promises.

In the United States, the Federal Trade Commission (FTC) has begun to focus on AI-driven mental-health tools, but even there the lag is stark. According to Global Privacy Watchlist, only a handful of AI health products have been examined by the FTC in the past two years, despite a flood of new releases.

Side-by-side analysis: launch speed vs regulatory review

That table makes it clear: developers can push code to production in a matter of days, while regulators are still drafting paperwork. The mismatch creates a risk environment where users may be exposed to untested algorithms.

Common regulatory gaps that expose users

• Misleading efficacy claims - apps often advertise "clinically proven" without peer-reviewed evidence.
• Data-privacy blind spots - many platforms store conversation logs in jurisdictions with weaker privacy laws.
• Algorithmic bias - AI models trained on limited demographics can misinterpret cultural expressions of distress.
• Lack of emergency protocols - few apps have built-in escalation paths for users reporting suicidal thoughts.
• Inadequate post-market surveillance - once an app is live, regulators rarely revisit its performance.
• Unclear liability - who is responsible if an AI misdiagnoses a condition?
• Regulatory overlap - TGA and ACCC sometimes issue contradictory guidance.
• International jurisdictional conflicts - many apps are hosted offshore, complicating enforcement.

These gaps are not just academic; they translate into real harm. In 2023, a Sydney-based trial of an AI-guided CBT app reported a 12% increase in dropout rates linked to poor user experience, a factor that the TGA later cited as a safety concern.

What state governments are doing - and why it matters

Some Australian states have taken matters into their own hands. Victoria's Department of Health released a “Digital Therapeutics Framework” in 2022 that requires any mental-health app used in public hospitals to undergo a rapid-review pilot. New South Wales, on the other hand, launched a Consumer Protection Unit dedicated to tech-related health complaints.

These initiatives are fair dinkum attempts to bridge the gap, but they are patchwork solutions. Without a national standard, a therapist in Perth might be using a fully vetted tool while a colleague in Brisbane is left with a version that hasn’t been checked for bias.

International perspective - U.S. AI mental health policy

Across the Pacific, the United States faces a similar conundrum. Federal agencies are still figuring out how to classify AI-driven therapy apps - are they medical devices, consumer products, or something new? The FTC has started issuing warning letters, but enforcement is sporadic. According to the Global Privacy Watchlist, only three AI mental-health tools have been the subject of a formal FTC investigation as of early 2024.

Meanwhile, individual states like California have introduced stricter data-privacy statutes that indirectly affect health apps. The patchwork of state-level rules creates a “regulatory lag on mental health apps” that mirrors Australia’s own federal-state split.

Practical steps for consumers

1. Check the TGA classification. If an app is listed as a medical device, it has passed a higher safety bar.
2. Read the privacy policy. Look for clear statements on data storage, sharing, and deletion.
3. Verify clinical evidence. Reputable apps will link to peer-reviewed studies or RCT results.
4. Beware of “instant cure” language. Any claim that a chatbot can replace a qualified therapist should raise a red flag.
5. Use reputable app stores. Platforms like the Google Play Store have begun flagging apps with deceptive health claims.
6. Ask your GP. Many general practitioners now maintain a list of approved digital mental-health tools.
7. Monitor your own data. If an app asks for unnecessary personal information, consider switching.

I've seen this play out when a friend in Melbourne signed up for a free AI therapy app that later sold her conversation data to a third-party advertising network. The experience left her reluctant to try any digital solution again.

What policymakers can do - a road map

• Adopt a risk-based classification. Treat AI therapy apps as a distinct category, separate from traditional medical devices.
• Mandate pre-launch safety testing. Require independent algorithm audits before an app can be marketed.
• Introduce fast-track review pathways. A 30-day provisional approval for low-risk apps would align with development cycles.
• Require transparent efficacy reporting. Companies must publish trial data in a publicly accessible registry.
• Standardise emergency escalation protocols. Any app detecting self-harm language should trigger a 24/7 helpline connection.
• Clarify liability. Legislate who is responsible when an AI recommendation leads to adverse outcomes.
• Coordinate federal-state efforts. A national digital-therapeutics board could harmonise standards across jurisdictions.
• Enforce data-privacy safeguards. Align app practices with the Australian Privacy Principles and impose heavy fines for breaches.
• Support public-sector pilots. Funding for government-run trials can generate robust evidence for safe AI tools.
• Educate clinicians. Offer training on how to evaluate and integrate AI-based tools into practice.

These steps would narrow the regulatory lag without stifling innovation. As the National Law Review predicts, AI regulation will tighten across the board in the next five years - we’d be better off getting ahead of the curve.

Key Takeaways

• AI therapy apps can launch in under a week.
• Federal reviews in Australia take 3-6 months.
• Regulatory gaps expose users to misleading claims.
• State pilots are patchwork, not a national solution.
• Clear policy steps can align safety with speed.

FAQ

Q: Why do AI therapy apps launch so quickly?

A: Developers use cloud platforms and continuous integration pipelines, allowing code to move from prototype to public release in days. The low cost of distribution and the ability to update remotely mean there’s little technical barrier to rapid launch.

Q: What does the TGA consider a medical device?

A: Per the TGA, software that provides diagnostic or therapeutic recommendations and influences clinical decisions is classified as a medical device. Apps that merely track mood without giving treatment advice fall outside that scope.

Q: How can I tell if an app’s claims are credible?

A: Look for links to peer-reviewed studies, check if the app is listed on the TGA’s medical device register, and watch out for vague phrases like “clinically proven” without citations. Independent reviews and professional endorsements also add credibility.

Q: What regulatory changes are expected in the next five years?

A: According to the National Law Review, governments will introduce risk-based classifications for AI health tools, fast-track approval pathways, and mandatory algorithm audits. Both Australia and the United States are moving toward tighter oversight as AI becomes more embedded in care.

Q: Where can I find a list of approved mental-health apps?

A: The TGA publishes a searchable register of approved medical-device software. State health departments, such as Victoria’s Digital Therapeutics Framework, also maintain curated lists of vetted apps for public use.