'My whole story was erased': Paris residents speak out as duplicate image removal sweeps community archives

Hundreds of photographs are gone. Not lost to a server crash or a flood, but quietly deleted by software designed to clean up digital clutter — and residents across Paris say nobody warned them it would take their memories with it.

The issue centres on the growing use of automated duplicate-image-removal tools by several of the city's community archive platforms and municipal digital projects. When the same photograph appears twice in a shared database, the software flags one copy for deletion. In practice, the algorithm cannot distinguish between a true duplicate and two different users who independently uploaded the same historically significant image. The result: originals vanish. Metadata — the date, the caption, the name of the photographer — vanishes with them.

The problem is surfacing now because of the scale of Paris 2024 Olympics legacy digitisation work. Since early 2025, the Mairie de Paris has been migrating thousands of neighbourhood photographs into a consolidated archive under the Grand Paris Mémoire Collective programme, a project linked to the broader Seine urban regeneration corridor. That migration triggered automated deduplication at a volume the system was never originally stress-tested for.

Archives emptied, communities blindsided

The Bibliothèque de l'Arsenal, on the Boulevard Morland in the 4th arrondissement, hosts one of the affected sub-archives. Staff there have been fielding complaints since March. Residents of the 18th and 19th arrondissements — neighbourhoods where community-led photography projects have documented banlieue life for over a decade — say entire folders of images submitted to local associations like the Collectif Barbès Mémoire have been reduced to single representatives, erasing context and duplicated effort that was, in many cases, deliberate.

One resident of La Chapelle, who submitted more than 200 photographs documenting the Rue Marx Dormoy street market between 2019 and 2023, described opening the archive in June to find fewer than 30 images remaining. The photographs that survived were not the highest-quality ones, nor the most recent — they were simply whichever copy the algorithm processed last. Personal captions written in French and in Kabyle were gone.

At the Maison des Associations in the 20th arrondissement, volunteers running a Belleville neighbourhood oral history project say they lost access to paired image-and-audio records. The software deleted what it read as duplicate JPEGs without recognising that each copy was linked to a distinct audio testimony stored separately. Rebuilding those links manually, they say, could take the rest of 2026.

Scale of the problem, and what comes next

Quantifying the damage is difficult because the deletions happened silently over several months. The Grand Paris Mémoire Collective programme launched its digitisation phase with a stated goal of consolidating more than 1.2 million images by the end of 2026, according to documentation published by the Mairie de Paris in January of this year. Administrators have not yet released a figure for how many images were removed by deduplication during the migration.

Community archivists familiar with similar transitions point to comparable cases in Lyon and Brussels, where municipal digitisation programmes adopted deduplication without community consultation and later had to establish manual review committees to assess what was recoverable. In Lyon's case, recovery took 14 months.

For Paris residents who still have local copies of their original uploads, the advice from digital preservation specialists is to act before the next scheduled migration window, currently set for September 2026. The Bibliothèque Publique d'Information at the Centre Pompidou, which offers free digital literacy workshops on the first Saturday of each month, has already added a session specifically on archival backup practices in response to the complaints.

Longer term, associations across the 18th, 19th and 20th arrondissements are asking the Mairie to suspend automated deduplication on community-submitted collections until a human review layer is built into the process. Whether city administrators will move fast enough to prevent another round of deletions before September is the question now hanging over every community archivist who still has something left to lose.