A Temu order containing seeds disguised as a “lucky bamboo plant kit.” A Shein parcel with what turned out to be unidentified bulbs tucked into the packing. An eBay purchase of “vintage gardening tools” that arrived caked in soil from a UK garden. Every week, DAFF’s mail centre operations intercept material that, ten years ago, simply wouldn’t have entered the country at this volume — because individual consumers weren’t directly importing it.

The shift from container-dominated trade to parcel-dominated consumer flow has fundamentally changed Australia’s biosecurity risk profile, and the inspection model is struggling to keep up. Inbound international parcel volume crossed 280 million units in calendar 2025, up from roughly 95 million in 2019. That’s a 195% increase in six years. Inspection resourcing has not grown at that rate. Nothing close.

Where the risk concentrates

DAFF’s mail risk profiling tells a fairly consistent story. Plant-related material (seeds, bulbs, dried herbs, plant-based food products) accounts for the largest single category of interceptions. Animal-derived products — meat snacks, dairy items, animal-fibre clothing with insect contamination — are a steady second. Wooden items including bamboo, untreated timber crafts, and some furniture components round out the high-frequency hit list.

The problem isn’t that DAFF doesn’t know what to look for. The problem is volume-to-resource mismatch. Random inspection rates at the Sydney Gateway Facility have dropped to under 8% of parcels in some weeks, down from 15–18% pre-pandemic. The trade-off is targeting based on declared contents and risk profiling — but declared contents are often wrong, deliberately or otherwise.

Reuters reported in January that Chinese e-commerce platforms had been the source of over 70% of high-risk plant material interceptions in calendar 2025. That’s not a surprising statistic given platform-share dynamics, but it does point to where targeted intervention should focus.

Detector dog programs and X-ray

The biosecurity detector dog program has been one of the genuine success stories of the past five years. The teams at Sydney, Melbourne, Brisbane, and Perth gateway facilities have detection rates that consistently outperform random inspection by an order of magnitude on the targeted categories. DAFF expanded the program in 2024 with eight additional teams, and there’s a strong case for further expansion.

X-ray screening remains the volume backbone. The newer dual-energy systems with auto-detection algorithms are noticeably better at flagging organic matter than the equipment they replaced, but they’re still operator-dependent. The throughput is constrained by how fast a trained reader can process imagery, and that’s where machine vision tools have started to help materially. DAFF has piloted automated image classification systems in partnership with several Australian AI vendors, including Team400, with reported sensitivity improvements on plant-material detection in test datasets.

The pilot results that DAFF has shared publicly suggest false-positive rates remain too high for fully automated decisioning, but as a triage layer the technology is clearly useful. The Inspector-General of Biosecurity flagged in their December 2025 review that machine vision deployment should accelerate, with appropriate validation governance.

Platform-level intervention

The most leveraged intervention point isn’t at the Australian border. It’s at the e-commerce platform listing stage. DAFF and the ACCC have done some work with Amazon, eBay, and Australia Post-affiliated platforms to flag and remove listings that promote biosecurity-prohibited items. That cooperation has been variable — generally good with Amazon, harder with platforms that don’t have an Australian corporate presence to engage.

The Department of Home Affairs and DAFF jointly proposed an expanded notification regime in late 2025 that would require platforms over a certain size to demonstrate due diligence on biosecurity-prohibited categories. The legislative pathway for this isn’t clear yet, and there’s reasonable industry concern about extraterritorial enforcement. But the policy logic is sound: you can either inspect 280 million parcels at the border, or you can stop a meaningful fraction from being shipped in the first place. The latter is cheaper.

What declared content data tells us

I’ve seen some of the aggregated declared-content analysis from the parcel post stream. The patterns are revealing. There’s a long tail of mislabelled “sample,” “gift,” and “documents” declarations covering biosecurity-relevant items. There’s also a clear seasonality — seed and bulb interceptions spike in February–April, which aligns with northern hemisphere garden ordering patterns being shipped to Australian buyers.

The data quality on declared values, contents, and origin is genuinely poor. ABF and DAFF have been talking about a unified declarations data model for years, and progress is slow. Better data wouldn’t solve the inspection capacity problem, but it would let targeting do more work.

What I’d watch over the next 12 months

The Inspector-General of Biosecurity has another scheduled review of mail and parcel risk management due in late 2026. That report, more than any single policy announcement, will tell us whether the system is keeping pace.

The other thing I’d watch is the cost-recovery debate. The mail screening program is partly funded through general appropriation, partly through container-level cost recovery on commercial trade. There’s a sensible argument that high-volume e-commerce platforms should bear more of the screening cost directly, given that their business model is generating the volume. The Productivity Commission has touched on this in passing but hasn’t taken it on properly. They probably should.

Biosecurity is one of those policy areas where the cost of failure is so much larger than the cost of prevention that the economics shouldn’t be controversial. The volume growth in parcel trade is real, persistent, and getting faster. The inspection system is in a genuine race.