'My Face Was Gone': Tokyo Residents Speak Out on the Spread of Duplicate and AI-Replaced Images Online

A 34-year-old ceramics instructor from Koenji discovered it by accident last autumn: a product listing on a major e-commerce platform was using a photograph of her hands — lifted from her personal website — to sell someone else's pottery from a shop she had never heard of. The image was hers. The credit was not. This is what the debate over duplicate image replacement looks like at street level in Tokyo.

The problem has sharpened considerably in 2026. Generative AI tools capable of seamlessly swapping faces, textures, and backgrounds have become cheap and accessible, and the volume of images circulating without their subjects' knowledge has grown faster than existing legal frameworks can track. For Tokyo residents already navigating a city dense with surveillance cameras, social media exposure, and inbound tourism that produces millions of photographs daily, the issue feels both abstract and acutely personal.

A Ward-by-Ward Problem

In Shimokitazawa, a neighbourhood known for its vintage clothing shops and live music venues, small business owners say the issue cuts directly into their livelihoods. One textile seller on Ichibangai shopping street found that product images from her online store had been copied and re-used on at least three separate overseas retail sites, with the original backgrounds digitally replaced to remove identifying features of her shop. She reported the incidents to the Tokyo Metropolitan Police Department's cybercrime consultation window, a service that logged more than 11,000 image-related complaints in fiscal year 2024, according to figures the department has previously published.

The NPO Digital Rights Tokyo, which operates a counselling desk in Shinjuku-ku near Okubo Station, says inquiries about unauthorised image duplication and AI-generated likeness replacement rose sharply in the first quarter of 2026 compared to the same period the year before. The organisation helps affected residents file takedown requests and navigate the Provider Liability Limitation Act, Japan's primary legal instrument for pursuing platform operators over harmful content. Staff there describe a pattern: the most common victims are freelancers, small retailers, and women whose social media profiles attract scraping bots.

Harajuku's Takeshita Street, a perennial draw for both domestic visitors and the roughly 36 million tourists who visited Japan in 2025, generates an enormous volume of street photography daily. Residents and shopkeepers along that strip report a secondary frustration: photographs taken of them or their storefronts by tourists are sometimes fed into AI tools that strip out faces, replace backgrounds, or clone the image into a synthetic scene with no resemblance to reality — yet bearing enough detail to mislead viewers about the original setting.

What the Law Can and Cannot Do

Japan's Act on the Protection of Personal Information, revised most recently in 2022, offers some protections around biometric data, but legal scholars at institutions including Meiji University in Chiyoda-ku have noted in public forums that the statute was not written with AI image manipulation specifically in mind. Filing a civil claim is expensive and slow. Platform takedown processes, while legally mandated to move quickly once a valid complaint is filed, in practice can take weeks when the infringing content sits on overseas servers.

The Cultural Agency, which sits within the national government structure, has been consulting on how copyright interacts with AI training data since 2023, but has yet to publish binding guidance specifically addressing the rights of individuals whose images are duplicated and altered without consent. A working group report had been expected in spring 2026; as of early July it had not appeared publicly.

For those already affected, Digital Rights Tokyo recommends three immediate steps: document the infringing URL with a timestamped screenshot, submit a report through the platform's official abuse portal using the specific language of Japan's Provider Liability Limitation Act, and — if the image appears on a Japanese-registered service — file a parallel notification with the Telecommunications Bureau of the Ministry of Internal Affairs and Communications. None of these routes is fast. But residents who have gone through the process say that filing through multiple channels simultaneously tends to accelerate results. The ceramics instructor from Koenji had her images removed from two of three offending listings within six weeks of filing. The third, hosted on a server registered in Eastern Europe, remains live.