'My Story Got Erased': New Yorkers Speak Out on Duplicate Image Replacements Stripping Community Archives

Denise Okafor has a photograph of her grandmother standing in front of the old Associated supermarket on Tremont Avenue in the Bronx, taken sometime in the early 1990s. She submitted it two years ago to a neighborhood digitization project run through the Bronx Documentary Center. Last spring, she tried to find it again. It was gone — replaced, she was told, by a system that flagged it as a duplicate of another image already in the archive.

Okafor is not alone. Across the city, community members who have contributed personal photographs, scanned documents and neighborhood snapshots to public archives and municipal digitization programs are discovering that automated deduplication tools — software designed to strip out repeated or near-identical images — have quietly removed materials they believed were permanently preserved. The issue has surfaced sharply in 2026 as the city expands its digital infrastructure ahead of the FIFA World Cup, with multiple agencies accelerating projects to catalog New York's public spaces and neighborhoods.

When Automation Meets Community Memory

The problem cuts across borough lines. At the Brooklyn Public Library's Community Archives unit on Grand Army Plaza, staff have fielded a growing number of complaints from residents whose submissions to the library's Flatbush and Crown Heights neighborhood history collections were overwritten after a 2025 software upgrade expanded automated deduplication across the system. The library confirmed in a March 2026 public notice that the upgrade affected materials submitted between January 2023 and October 2025, though the library has not released a count of how many individual items were affected.

At the Queens Memory Project, based out of the Queens Public Library in Jamaica, coordinators have been manually auditing submissions since February after community members flagged missing materials. The project, which has collected more than 14,000 items since its founding in 2010, uses a combination of automated and human review — a safeguard that appears to have limited the damage compared to systems that rely more heavily on automation.

The distinction matters. Deduplication software typically matches images by pixel similarity or file hash, which means two photographs of the same street corner taken by different people on different days can be treated as redundant. For community archivists, that logic misses the point entirely. The value in a neighborhood photograph is often contextual: who took it, when, and why.

What Residents Are Demanding

Community groups in Washington Heights and Sunset Park have begun organizing around the issue, pushing for mandatory human review of any automated deletion from publicly funded archives. El Barrio's Artspace PS109, which hosts cultural preservation programming on East 104th Street in East Harlem, held a public meeting in June where residents raised the deduplication problem alongside broader concerns about who controls the digital image of their communities.

The timing adds pressure. The city is spending heavily on digital and physical infrastructure for the 2026 World Cup, with matches scheduled at MetLife Stadium in New Jersey and fan zones planned across Manhattan. That investment has accelerated digitization projects at multiple agencies, increasing the volume of images being processed — and the likelihood of automated systems making consequential decisions at scale.

City Council Member Shaun Abreu, whose district covers parts of Washington Heights and Inwood, held a constituent town hall on digital preservation in May. No formal legislation has been introduced yet, but community advocates have submitted testimony to the Council's Cultural Affairs committee calling for an audit of deduplication practices across city-funded archive programs.

For those who have lost materials, the practical path is limited. The Bronx Documentary Center advises residents to resubmit items with detailed metadata — contributor name, date, location, a written description — which reduces the likelihood of an automated system flagging the image as redundant. The Brooklyn Public Library has said it is reviewing whether any removed items can be recovered from backup servers, with a preliminary assessment due before the end of July 2026.

Okafor says she still has the original photograph at home. She is considering resubmitting it, but she is not sure she trusts the system. Her grandmother's image of Tremont Avenue in the 1990s is not a duplicate of anything. It is the only copy of what she remembers.