'My Grandmother's Face Was Erased': SF Residents Speak Out on Duplicate Image Removal Policies

Maria Solis kept a digital archive of more than 400 photographs on a community platform managed through the Mission District's nonprofit network. Last spring, she logged in to find 60 of them gone — flagged and auto-deleted by a duplicate-detection algorithm that the platform had quietly enabled in March 2026. Among the missing: a scan of her late grandmother taken in Guadalajara in 1971, the only copy she had digitized. The original had been destroyed in a house fire on Folsom Street years earlier.

Solis is not alone. Across San Francisco, residents using city-affiliated digital archives, nonprofit photo repositories, and community history platforms are reporting that automated duplicate-image-removal systems — tools originally designed to reduce storage costs — are sweeping out photographs that cannot be recovered. The issue has moved from a technical inconvenience into something that community advocates describe as a quiet cultural erasure, hitting hardest in neighborhoods where residents have less redundancy in how they store personal and historical records.

A Problem Rooted in Shrinking Budgets and Automated Shortcuts

The timing matters. Since late 2024, a wave of cost-cutting across San Francisco's nonprofit and civic-tech infrastructure — accelerated by the withdrawal of several federal digital-equity grants — has pushed organizations to adopt cheaper, more automated content-management tools. Many of these tools use perceptual hashing, a technique that identifies visually similar images and flags duplicates for deletion. The problem is that the threshold for what counts as a "duplicate" is often set aggressively, and images that are slightly different scans of the same physical photograph — or two photos taken seconds apart of the same person — can be treated as redundant files.

The San Francisco Public Library's Digital Collections program, based at the main branch on Larkin Street in Civic Center, has so far kept human review in its workflow. But smaller organizations operating on thinner margins have not had that luxury. At least three Mission-based community archives and one Tenderloin neighborhood history project have adopted third-party tools that include automated duplicate removal as a default-on feature, according to people familiar with those organizations' operations. The Daily San Francisco is not naming those organizations because the extent of their data losses has not been independently verified and they have not been given opportunity to respond.

The broader digital storage market offers some context for why this is happening now. Cloud storage costs, while historically declining, have stabilized or edged upward since 2024 as major providers restructured pricing tiers. For a nonprofit running a photo archive of 100,000 images, the difference between storing every file and aggressively deduplicating can translate to hundreds of dollars per month — a real pressure point for organizations operating on program budgets that often run under $500,000 annually.

Chinatown, the Excelsior, and Demands for Human Oversight

Community historians in the Excelsior District and Chinatown have been among the most vocal about the consequences. The Chinatown Community Development Center, headquartered on Powell Street, has fielded calls from residents asking how to recover images removed from platforms the organization does not directly control. In the Excelsior, a neighborhood oral-history project connected to San Francisco State University's College of Ethnic Studies lost a tranche of digitized family portraits last February after a contractor updated the project's content-management system without flagging the new deduplication settings to project staff.

Advocates are now pushing for what some are calling a "human-in-the-loop" standard — a requirement that any automated deletion of user-submitted images go through a 30-day quarantine period with notification to the uploader before permanent removal. Several cities, including New York, have begun exploring similar notification mandates for municipal digital platforms, though no binding policy is yet in place anywhere in the United States as of this writing.

For San Franciscans who have already lost images, the practical options are limited but not zero. The Internet Archive, based on Funston Avenue in the Richmond District, accepts community submissions and maintains its own preservation protocols separate from commercial cloud tools. The San Francisco Public Library's Digital Preservation Lab on Larkin Street offers digitization assistance and uses formats designed for long-term storage. Residents who believe images were wrongly removed from a platform should file a written request for restoration within 90 days — many terms of service include a recovery window that users rarely know exists.