How Canberra's Public Sector Archives Ended Up Drowning in Duplicate Images — and What Comes Next

The ACT government's digital asset problem did not appear overnight. Across dozens of agencies housed in everything from the Civic-based Nara Centre to offices along Northbourne Avenue, public servants have spent the better part of a decade uploading, re-uploading, and re-naming the same photographs, maps, and infographics — sometimes hundreds of times — as departments migrated from one content management system to the next without a coordinated handover process.

Duplicate image replacement — the systematic identification and removal of redundant visual assets from government and institutional digital libraries — has emerged as a practical priority in 2026 because storage costs and compliance obligations have finally caught up with the mess. The Australian National Audit Office flagged digital asset governance as a recurring weakness in its broader ICT procurement reviews, and the ACT's own Digital Strategy, updated in late 2024, included explicit language about data duplication as a cost driver that agencies were expected to address.

How Canberra Got Here

The roots go back to roughly 2017 and 2018, when a wave of machinery-of-government changes reshuffled ACT directorates and left digital libraries stranded across incompatible platforms. The Transport Canberra and City Services directorate, for instance, absorbed functions from several predecessor bodies, inheriting image repositories that had never been reconciled. The same pattern repeated itself during the 2020 and 2021 COVID-19 digitisation push, when agencies scrambled to put public-facing content online quickly, often duplicating assets simply because staff could not locate originals in legacy systems.

At the Australian National University on Acton Peninsula, library and IT staff encountered a parallel problem. ANU's digital collections — spanning research imagery, campus photography used in communications, and archival scans — had grown across at least three separate platforms by 2023, with no single authoritative source for commonly used assets. The University of Canberra at Bruce faced similar friction as it expanded its online course delivery footprint, with course design teams in different faculties independently uploading the same lecture graphics and campus images into the Moodle learning environment and separate marketing repositories.

These are not minor housekeeping inconveniences. Storage on enterprise cloud platforms, depending on contract tier, runs roughly between $23 and $80 per terabyte per month for Australian government-grade infrastructure under the AWS and Microsoft Azure panels available through the Digital Transformation Agency's whole-of-government agreements. For large agencies carrying tens of thousands of redundant image files, that represents a tangible and recurring budget line.

The Compliance Pressure Sharpening the Urgency

Two regulatory threads have pushed duplicate image replacement from a nice-to-have to a near-term operational task. First, the ACT's Records Management Standard, administered through Archives ACT, requires that agencies maintain accurate and non-redundant records — a requirement that digital asset libraries have historically been allowed to quietly sidestep. Second, the federal government's updated AI use policy, which began shaping ACT public sector practice through 2025, requires that training data and image sets used in any AI-assisted communications tool be deduplicated and properly licensed, meaning agencies feeding images into automated content tools now face a direct compliance obligation.

For public servants working out of the Barton and Parkes precincts — where several major Commonwealth agencies maintain ACT-facing offices — the practical consequence has been a wave of internal audits since early 2026. Some agencies are running commercial deduplication software. Others are attempting manual reviews, which is labour-intensive and prone to error at scale.

The path forward varies by organisation, but the clearest model emerging in the ACT involves three steps: an automated scan using hash-matching tools to flag bitwise duplicates, a secondary review of near-duplicates using perceptual hashing software, and then a governed deletion process with sign-off from records managers. Archives ACT published updated guidance on digital asset disposal in March 2026 that agencies are now expected to align with. For institutions like ANU and UC, the process is less formally mandated but the reputational and operational incentives are real. Staff time, storage spend, and the growing use of AI content tools are making a clean library less of an aspiration and more of a baseline requirement.