How San Francisco's City Agencies Spent Years Paying for the Same Stock Photos Twice — and Sometimes Three Times

San Francisco's municipal government has been quietly grappling with a sprawling duplicate image problem — thousands of identical or near-identical photographs stored across separate servers, licensed multiple times from vendors, and embedded in overlapping city websites — that auditors and department technology officers say has compounded over at least a decade of siloed procurement decisions. The issue, which touches everything from the Department of Public Health's public-facing portals to the SF Planning Department's neighborhood project pages, is finally drawing formal attention as the city undertakes a wider digital infrastructure consolidation push in 2026.

The problem matters now because San Francisco is mid-way through a technology modernization effort that began in earnest after Mayor London Breed's administration committed, in her final budget cycle before the November 2025 election, to consolidating the city's fragmented digital assets under the Department of Technology's unified content management system. The new administration has inherited that project, and auditors reviewing the transition have found that duplicate image files — some licensed at anywhere from $400 to $1,200 per image from commercial stock providers — account for a measurable share of unnecessary annual licensing overhead. Resolving it is now a prerequisite for the broader migration.

How the Mess Accumulated

The roots of the problem stretch back to the early 2010s, when individual city departments began building their own websites largely independently. The SF Recreation and Parks Department, the Office of Economic and Workforce Development on Mission Street, and the Municipal Transportation Agency each contracted separately with web vendors who sourced their own image libraries. Nobody was cross-checking. By 2018, the city's main sfgov.org portal alone had undergone three separate redesigns, each of which imported legacy image folders without auditing what was already there.

BART and Muni's public communications arms compounded the pattern. Marketing materials produced for Muni's Van Ness Bus Rapid Transit corridor — which opened in phases between 2022 and 2023 — were photographed, archived, and then re-photographed by at least two separate contracted agencies, generating duplicate libraries that were both billed to the MTA. Neither set of files was tagged with enough metadata to make deduplication straightforward.

The city's Digital Services team, housed at 1 Dr. Carlton B. Goodlett Place in Civic Center, did not have enterprise-wide visibility into departmental image repositories until a 2024 citywide audit flagged redundancy as a systemic issue rather than an isolated one. That audit identified more than 40 city subdomains carrying overlapping visual assets, with some hero images — the wide-format photographs used at the top of web pages — appearing in as many as seven separate file directories.

What Deduplication Actually Requires

Fixing duplicate image libraries is not simply a matter of deleting files. Legal teams at the City Attorney's Office must first verify which licenses are still active, which images are in the public domain, and which were produced under work-for-hire contracts that vest copyright in the city outright. That legal triage alone, for a repository estimated to contain tens of thousands of assets across all departments, is expected to take most of 2026 to complete.

The SF Public Library's digital archives team, based at the main branch on Larkin Street, has been brought in as a consulting resource because librarians there have spent years developing metadata standards and deduplication workflows for historical photograph collections. Their approach — using perceptual hash matching software to identify visually identical images regardless of filename — is now being piloted on the Department of Public Health's asset server as a proof of concept.

For city residents, the practical consequence of cleaning this up is modest but real. Faster-loading city web pages, lower annual licensing costs, and a single searchable asset library that departments can share instead of re-purchasing the same photograph of, say, the Ferry Building or Dolores Park for the fourth time. Technology officers say a completed deduplication would allow the city to renegotiate its stock image contracts from a position of knowing exactly what it actually needs — rather than buying in bulk to cover gaps nobody has fully mapped. The consolidation project's target completion date is the second quarter of 2027.