Singapore's War on Duplicate Images: How the City-State Stacks Up Against Tokyo and London

Singapore's Infocomm Media Development Authority confirmed earlier this year that duplicate and near-duplicate image proliferation had become a measurable problem on government-linked digital portals, affecting everything from HDB property listings on the Housing & Development Board's resale flat marketplace to heritage archive entries maintained by the National Library Board. The scale of the issue — redundant visuals clogging search results, inflating storage costs and undermining public trust in official information — has pushed Singapore to accelerate detection and removal protocols that other major cities are only beginning to draft.

The timing matters. Generative AI tools capable of producing near-identical image variants at scale became widely accessible in 2023 and 2024. By mid-2025, duplicate image complaints on Singapore government portals had risen sharply enough that the Government Technology Agency, GovTech, incorporated deduplication checkpoints into its whole-of-government data quality framework. The concern is not purely cosmetic. When a Tampines flat and a Jurong West flat share identical listing photographs — a problem documented in the HDB resale portal — buyers lose confidence in the data's reliability, and that erodes trust in one of the city-state's most socially sensitive institutions.

What Singapore Is Actually Doing

GovTech's Digital Infrastructure office has been running a perceptual hashing pipeline — a technique that generates a compact fingerprint from each image so near-identical copies can be flagged even after minor cropping or colour adjustment — across several public-facing platforms since the fourth quarter of 2025. The National Heritage Board, which manages digitised collections at institutions including the Asian Civilisations Museum on Empress Place and the National Museum of Singapore on Stamford Road, began a parallel deduplication audit in January 2026, working through roughly 1.2 million digitised assets in its online repository.

Private sector pressure is also mounting. PropertyGuru, which competes directly with HDB's own listing tools, announced a machine-learning image deduplication layer in March 2026, citing user-experience data showing that repeated property photographs were among the top three reasons users abandoned search sessions. The platform did not disclose the volume of duplicates detected, but the public acknowledgement itself signals how broadly the problem has spread beyond government systems.

How Singapore Compares With Tokyo and London

Tokyo's approach has been characterised by institutional caution. Japan's Digital Agency, established in September 2021, has prioritised data interoperability over aggressive content moderation on public portals, meaning duplicate image removal remains largely manual and siloed within individual ministries. The National Diet Library's digital archive — one of Asia's largest — still relies primarily on metadata flags rather than automated visual comparison, according to publicly available documentation on the agency's website.

London sits somewhere between the two. Transport for London's open data portal and the Greater London Authority's planning portal both implemented basic duplicate-detection scripts following a 2024 audit by the UK government's Central Digital and Data Office. However, the scope was narrower — focused on planning application photographs rather than broader public information assets — and rollout remained confined to a handful of pilot boroughs through the first half of 2026.

Singapore's advantage is structural. Because GovTech operates as a central shared-services agency serving all ministries, a single deduplication standard can cascade across dozens of platforms simultaneously rather than requiring individual agencies to commission their own solutions. Tokyo and London both operate in more fragmented digital governance environments, where a borough council or a standalone ministry may use entirely different content management systems with no common image registry.

For members of the public, the practical upshot in Singapore is already visible. Residents searching for rental listings in Clementi or Bedok on government and government-linked portals are increasingly unlikely to see the same stock photograph appearing under three different addresses — a problem that was common enough eighteen months ago to generate complaint threads on online forums. GovTech has indicated it intends to extend deduplication standards to statutory board websites by the end of 2026, which would bring agencies including the Urban Redevelopment Authority and the Land Transport Authority into the same framework. Whether the private market moves as quickly is a separate question — but with PropertyGuru already acting, the pressure on competitors is real.