Hong Kong's War on Fake Images: What Officials, Experts and Key Figures Are Saying

Hong Kong authorities and digital media specialists are stepping up calls for standardised protocols to detect and remove duplicate or manipulated images from public-facing platforms, as concerns mount that unchecked image recycling is undermining trust in government communications, news media and commercial advertising across the city.

The issue has gained urgency in 2026, with the Office of the Government Chief Information Officer — headquartered on Lower Albert Road, Central — circulating internal guidance this year on image integrity standards for departments that publish visual content online. While the guidance has not been made fully public, its existence reflects a broader institutional reckoning with how quickly fabricated or re-used imagery spreads through platforms still accessible in Hong Kong, including Telegram channels and locally hosted news aggregators.

Why It Matters Now

The timing is not coincidental. The global spread of AI-generated imagery has accelerated sharply since 2024, and Hong Kong sits at a particular crossroads: a common law jurisdiction with a free press tradition, but also one where the Article 23 legislation enacted in March 2024 has expanded the legal definition of content that can be deemed seditious or misleading. Legal scholars at the University of Hong Kong's Faculty of Law on Pokfulam Road have noted, in published commentaries, that manipulated visual content could in certain circumstances attract scrutiny under sedition provisions — a point that has made editors and platform operators notably cautious.

The Hong Kong Press Freedom Index, published annually by Hong Kong Public Opinion Research Institute, recorded a media environment score of 26.5 out of 100 in its most recent report — figures that reflect the backdrop against which image-verification debates are unfolding. Against that environment, the practical question of who bears responsibility for catching duplicate or synthetic images before publication has become both a technical and a legal one.

The Hong Kong Internet Registration Corporation Limited, known as HKIRC, which manages the .hk domain registry from its offices in Kowloon Bay, has been in discussions with industry groups about extending its existing content-integrity pilot — originally scoped for phishing detection — to cover visual metadata verification. No formal programme has been announced.

What Experts Are Recommending

Digital forensics professionals advising newsrooms along Wanchai's newspaper corridor and at Cyberport in Pok Fu Lam broadly agree on a short list of technical interventions: perceptual hashing tools that fingerprint images at upload, reverse-image search integration into content management systems, and mandatory provenance tagging using the C2PA standard — a specification backed by major camera manufacturers and platforms globally since 2021.

The Hong Kong Productivity Council, which operates training and technology-transfer programmes from its Kowloon Tong campus, ran a workshop series on AI content authentication in the first quarter of 2026, drawing participants from advertising agencies, broadcasters and at least two government departments. Attendees were shown demonstrations comparing images flagged by automated hash-matching against those cleared by human review, with error rates in automated systems running at roughly 3 to 7 percent depending on compression quality — a figure that practitioners said was low enough for first-pass filtering but not sufficient to replace editorial judgment.

Smaller outfits — independent news sites, district council communications offices, e-commerce sellers on platforms like HKTVmall — face a steeper climb. Licensing enterprise-grade verification tools can cost upwards of HK$8,000 per month, putting them out of reach for many local operators without subsidised access.

The next concrete milestone will be a review session scheduled by the Communications Authority for the third quarter of 2026, at which broadcasters and online news platforms licensed under the Broadcasting Ordinance are expected to present their current image-verification workflows. Whether those sessions produce binding requirements or remain advisory will depend heavily on how the Authority weighs the competing pressures of content integrity, regulatory scope and the operating costs that smaller licensed operators say they cannot easily absorb.