Canberra's Duplicate Image Problem: What Happens Next and the Key Decisions Ahead

Canberra's public sector is sitting on a growing crisis it rarely talks about publicly: digital archives bloated with duplicate images, redundant scans, and uncatalogued photographs that are costing agencies real money in storage, slowing records retrieval, and — in some cases — creating compliance headaches under the Territory Records Act 2002. The ACT government and its major institutional partners must now decide how aggressively to act, and which tools to trust.

The timing matters. The federal government's AI in Government framework, updated earlier this year, has opened the door for Commonwealth agencies headquartered in Canberra — from the Australian Public Service Commission on Constitution Avenue to Services Australia's Tuggeranong offices — to deploy automated deduplication tools at scale. That's put the question squarely on the desk of IT and records managers across the capital: automate now, or wait for clearer standards?

Where the Pressure Is Building

At the Australian National University in Acton, the archives and special collections team holds hundreds of thousands of digitised images spanning decades of research photography, aerial surveys, and institutional records. Managing duplicates across those collections is already a recognised operational challenge — the kind that consumes archivists' hours and inflates cloud storage bills. ANU's Scholarly Information Services division has flagged digital asset management as a priority in its current planning cycle, though no specific deduplication program has been publicly announced.

The ACT's own records authority — Archives ACT, based in the Dickson area — is the body responsible for advising government directorates on records management standards. Under the Territory Records Act, agencies have legal obligations around the retention and accessibility of official records, and duplicate image files complicate both. A single misidentified duplicate deleted from a ministerial correspondence folder could constitute a records breach. The stakes, archivists will tell you, are not trivial.

Meanwhile, the ACT Health directorate, which operates from its Canberra Health Services administrative hub near Garran, manages vast quantities of medical imaging data. Duplicate DICOM files — the standard format for radiology scans — are a known inefficiency in hospital IT systems nationally. The Australia Institute of Health and Welfare estimated in its 2025 digital health infrastructure report that duplicate or redundant imaging data contributes meaningfully to storage cost overruns across state and territory health systems, though ACT-specific figures have not been publicly released.

The Decisions That Will Define the Next 12 Months

Three choices are converging. First, agencies must decide whether to run deduplication algorithmically — using hash-matching or perceptual hashing tools that compare images at a pixel level — or rely on manual review workflows. Automated tools are faster and cheaper but carry a non-zero error rate; wrongly flagged originals get deleted alongside genuine duplicates. For records subject to the Territory Records Act, that risk needs sign-off from Archives ACT before any bulk deletion proceeds.

Second, procurement decisions loom. Several Canberra-based government ICT teams are currently mid-tender for digital asset management platforms, with contracts expected to be awarded before the end of the 2026-27 financial year. Which platforms get selected will largely determine what deduplication capabilities agencies have access to — some enterprise systems include automated deduplication as standard; others treat it as a paid add-on.

Third, there's the question of governance. Who reviews the results of a deduplication sweep before files are permanently deleted? At the University of Canberra's Bruce campus, the library and learning services team uses a tiered review model — automated flagging, human spot-check, then deletion — that records managers at other institutions have pointed to as a workable model. It's slower than full automation but dramatically reduces the risk of irreversible errors.

For public servants and researchers working across Canberra's civic institutions, the practical upshot is straightforward: expect new internal policies on image file naming, storage location, and periodic audits to start appearing in the second half of 2026. Agencies that move early, get governance right, and pick the correct tools stand to cut storage costs and meet their records obligations more cleanly. Those that delay may find the problem compounds — and that regulators, eventually, come looking.