Dozens of Zurich-based photographers, small-business owners and cultural organisers say they discovered only after the fact that automated duplicate-image detection tools had quietly deleted photographs they believed were safely stored — in some cases wiping years of visual documentation without any warning or recovery option.
The problem centres on deduplication algorithms built into cloud storage platforms and content management systems. These tools scan uploaded files for visual similarities and remove what they flag as redundant copies. The logic is efficient enough in theory. In practice, residents say the systems cannot distinguish between intentionally duplicated backups and subtly distinct versions of the same scene — deleting both, or the wrong one, with no straightforward undo.
The timing matters. Zurich's housing shortage has pushed cultural and community organisations out of established city-centre venues and into newer districts where documentation of neighbourhood life has taken on fresh urgency. Groups recording the transformation of areas like Altstetten and the Langstrasse corridor say photographic records have become evidence in cantonal planning debates, cited in community submissions under Switzerland's direct-democracy initiative process. Losing images is not merely an aesthetic inconvenience.
What residents say happened
A community photography collective based at the Rote Fabrik cultural centre on Lake Zurich's western shore says it lost a batch of images from a 2024 documentation project covering the Escher-Wyss-Platz redevelopment. Members describe uploading what they believed were incremental backups across two separate accounts on the same platform. The deduplication engine treated the files as identical and retained only one set — then deleted that set when a storage-tier downgrade triggered a second automated clean-up. The collective was not contacted before either deletion.
In the Kreis 5 district, a small graphic-design studio near Pfingstweidstrasse reports losing a client archive of roughly 4,000 product images after migrating to a new content management system that ran deduplication on import. The studio had maintained master files and compressed web exports as separate folders, a standard professional workflow. The algorithm collapsed them into single entries, then retained only the compressed versions. Replacing the missing masters required reconstructing roughly six weeks of billable work.
At the Stadtbibliothek Zürich's Pestalozzi branch on Zähringerplatz, staff who run a digital-literacy programme for older residents say the issue comes up regularly in their drop-in sessions. Participants describe arriving with USB drives containing family photographs, uploading them to cloud services for safekeeping, and later finding that near-identical shots — sequential frames taken seconds apart, common on smartphones — had been silently merged or removed.
What the evidence shows
No Swiss-specific study has yet quantified the scale of unintended deletion from consumer deduplication. However, a 2025 report by the European Union Agency for Cybersecurity (ENISA) on cloud data integrity found that roughly 12 percent of surveyed small-to-medium enterprises across EU member states had experienced at least one unintended data loss event attributable to automated storage management in the preceding two years. Switzerland is not an EU member but its businesses use the same international platforms subject to those findings.
Swiss federal data-protection law, updated under the revised Federal Act on Data Protection that came into force in September 2023, requires data processors to implement appropriate technical safeguards against accidental loss. Whether automated deduplication without user notification constitutes a breach of that obligation is a question legal observers say has not yet been tested in a Swiss court.
The cantonal consumer protection office, the Kantonales Konsumentenforum, has not issued guidance specific to deduplication, though its general advisory materials cover cloud-storage risks.
Practical steps recommended by digital-preservation specialists include maintaining at least one local backup entirely separate from any cloud account, using file-naming conventions that embed version numbers or shooting timestamps, and explicitly checking the deduplication settings — often buried in account administration pages — before migrating large archives. For organisations whose images may be cited in official planning or legal processes, storing files in open formats such as TIFF alongside compressed exports adds a layer of protection that most deduplication engines are not configured to touch.
The Stadtbibliothek Zürich's digital-literacy sessions at the Pestalozzi branch run on the second and fourth Tuesday of each month, free of charge, and the programme coordinator confirmed the topic will be added to the autumn 2026 curriculum.