Duplicate and AI-generated images are flooding Singapore's digital platforms — and the agencies, academics and industry figures tasked with policing them are running out of patience. The Infocomm Media Development Authority has flagged image integrity as a priority enforcement area under its ongoing content advisory framework, with platforms operating under the Online Safety Act now required to demonstrate active moderation of manipulated visual content.
The urgency is real. Singapore's digital economy, valued at over S$106 billion according to the Government's Digital Economy Framework released in 2023, depends heavily on trusted visual information — from HDB resale flat listings on PropertyGuru to product images on Lazada and Shopee. When duplicate or synthetically generated images circulate unchecked, consumers lose money and platforms lose credibility.
What the Institutions Are Saying
Researchers at the National University of Singapore's School of Computing have been developing detection tools for the past two years, focused specifically on image hashing and perceptual similarity algorithms that can flag near-duplicate content even after cropping, colour-shifting or watermark removal. The work feeds into a broader conversation happening at the Singapore Management University's Sim Kee Boon Institute for Financial Economics, where analysts have noted that duplicate product images are disproportionately associated with counterfeit goods listings — a concern that Orchard Road retailers and the Singapore Tourism Board have both raised with the Singapore Retailers Association.
The Consumer Association of Singapore, based in Buona Vista, has received a rising number of complaints related to online listings showing images that do not match delivered goods. Without citing a precise breakdown it has not publicly released, the association has acknowledged the trend is concentrated in e-commerce categories including electronics and fashion — two sectors that dominate the Jurong East and Bugis shopping corridors.
At the government level, the Ministry of Digital Development and Information has pointed to the AI Verify framework — developed by IMDA and the AI Governance Advisory Council and launched in 2022 — as the existing mechanism for assessing trustworthiness in AI systems, including those that generate or manipulate images. Critics within the tech industry, however, have argued the framework is voluntary and carries no enforcement teeth for non-enterprise actors.
Industry Figures Push for Harder Standards
The debate intensified after reports emerged earlier this year that several property listing platforms were inadvertently hosting AI-generated room photographs — images that depicted flats in Toa Payoh and Tampines with layouts that do not correspond to any actual HDB floor plan. The Housing and Development Board confirmed it is working with listing platforms to establish image verification standards, though no mandatory rollout date has been announced.
Digital forensics professionals in Singapore point to PhotoDNA — a hashing technology used by Microsoft and licensed to large platforms — as a practical baseline solution already deployed against child safety image databases. Extending similar technology to cover commercial duplicate image abuse is technically feasible, specialists say, but requires a central image repository that Singapore's Personal Data Protection Act currently complicates.
Singapore Press Holdings' digital properties and Mediacorp have both updated internal image verification workflows in the past 18 months, cross-checking submitted visual content against reverse-image search tools before publication. Smaller independent media outlets along the Telok Ayer and Tanjong Pagar media cluster have fewer resources to do the same.
The practical advice from the current consensus is layered. Platforms should implement perceptual hashing at the point of upload, not after the fact. Businesses listing products or properties online are urged to embed verifiable metadata — including GPS coordinates and device identifiers — in original photographs before submission. And consumers are advised to use Google Lens or TinEye before completing any transaction that hinges on a single image. IMDA has said updated advisory guidelines for platforms on image moderation are expected before the end of 2026.