Findings and current scope
The Australian Therapeutic Goods Administration has completed its review of the conditional exclusion for certain digital mental health tools. Findings published on 19 October 2026 identify a mismatch between the exclusion criteria and increasingly complex products, including those using artificial intelligence. The update is a policy-development step, not a final regulatory amendment.
Not every mental-health or wellness app is a medical device. For relevant software, item 14E provides a conditional exclusion where the tool follows clinical practice guidelines established and widely accepted for use in Australia, referenced and displayed within the software for users to review. A guideline link in a user manual alone is insufficient. An excluded product is outside TGA regulation; exclusion is not evidence of TGA approval.
The review found that the criteria do not address intended purpose, user groups, output types, clinical claims or failure consequences. Higher-risk tools may therefore qualify. TGA also cautioned that information about the limited apparent use of this pathway was volunteered by stakeholders, and individual products were not assessed in detail. The findings inform policy development rather than establish market-wide compliance.
Clinical evidence and software behaviour
Consider a hypothetical anxiety-management app that adds AI-generated dialogue to a previously fixed set of exercises. Its cited guideline might remain unchanged while its outputs, degree of personalisation and influence on users change substantially. This scenario illustrates a product-development question, not a regulatory determination about a particular app.
Guideline alignment establishes an important basis for content. It does not, by itself, demonstrate that every implemented feature or generated response remains within the intended scope. Teams must still be able to explain what their software produces, how users are expected to interpret it, and how errors could affect subsequent decisions.
User expectations deserve equal attention. In TGA’s earlier survey, consumer respondents considered cost, reviews and ratings, developer identity and whether a tool works when choosing tools. Professional respondents placed greater emphasis on evidence, privacy and data protection. A boundary understood internally may not be equally apparent to the person using the product.
Consultation and preparation
TGA plans public consultation on regulatory options in 2026. As of 20 October, the reviewed update does not establish a final replacement rule or commencement date.
Manufacturers and suppliers can revisit the product’s boundaries before assuming that an earlier assessment covers new functionality:
- Reconcile intended use and clinical claims across specifications, interfaces and promotional materials.
- Map intended users, outputs and plausible consequences of incorrect information.
- Assess whether AI additions change the output boundary and whether existing evidence and verification cover those changes.
These are editorial recommendations for product and regulatory preparation, not newly imposed TGA requirements. The central issue is traceability between clinical rationale and implemented behaviour. A reference identifies the underlying guideline; the product team must be able to explain what the software does with it.
Sources
- Primary official sourceReview of the exclusion for certain digital mental health tools (opens in a new tab)Therapeutic Goods Administration
- Supporting sourceUnderstanding the digital mental health tool software conditional exclusion (opens in a new tab)Therapeutic Goods Administration
- Supporting sourceDigital Mental Health Tools User Survey (opens in a new tab)Therapeutic Goods Administration
- Supporting sourceMedical devices reforms: software regulation (opens in a new tab)Therapeutic Goods Administration