Webinar: Medical Devices

Device Regulation Tightens

Australia's medical device landscape is undergoing its most significant regulatory tightening in years. On 5 February 2026 the Therapeutic Goods Administration released updated guidance on artificial intelligence and medical device software, making clear that any software intended for diagnosis, prevention, monitoring, prediction, prognosis, treatment or alleviation of disease qualifies as a regulated medical device based solely on its purpose, not the underlying technology.

This places developers of AI radiology analysers, predictive deterioration apps or generative diagnostic chatbots squarely within the software-as-a-medical-device rules, requiring Australian Register of Therapeutic Goods inclusion, risk-based classification, clinical evidence and ongoing change control.

Barely six weeks later, from 21 March 2026, every public, private and day hospital in the country enters Stage 1 of mandatory adverse-event reporting to the TGA for incidents involving high-risk devices that cause death, serious injury or serious deterioration. The requirement, phased through 2030, closes surveillance gaps highlighted by earlier inquiries into implantable devices.

The next milestone arrives on 1 July 2026. Sponsors of Class III and Class IIb devices – including most implants – must begin placing Unique Device Identification carriers on labels and higher packaging levels and submit UDI records to the Australian UDI Database within 30 days of supplying the device in Australia. Legacy EU-certified high-risk devices receive staggered extensions into 2028, but new supply faces the hard deadline; full direct marking and stock relabelling stretch to 2030.

Parallel pressure comes from the 1 July 2026 cutoff for sponsors of existing ARTG entries containing non-viable animal tissues or cells to submit reclassification applications under amended rules; continued supply is permitted only until decisions are finalised. Meanwhile the TGA has broadened acceptance of certain overseas conformity assessments, including FDA pathways paired with MDSAP, to reduce duplication.

The cumulative effect hits multinational manufacturers hardest, as Australia represents a modest slice of global sales yet now demands its own database infrastructure, barcode systems and evidence packages. Hospitals gain tools for precise patient-record linkage and rapid recalls, while patients stand to benefit from earlier detection of safety signals. The less obvious trade-off is regulatory agility versus innovation pace: by folding AI into familiar risk categories rather than inventing a bespoke regime, the TGA avoids over-burden but still insists on transparent clinical validation that synthetic data alone cannot fully satisfy.