São Paulo's Digital Archive Crisis: The Key Decisions Ahead on Duplicate Image Replacement

São Paulo's municipal and private digital infrastructure is sitting on a growing backlog of duplicated image files, a technically unglamorous but operationally serious problem that is forcing decisions from City Hall on Paulista Avenue all the way to the tech startups clustered in Vila Olímpia and Faria Lima. The core question — what system replaces duplicates, who pays for the cleanup, and which records get priority — is now being actively debated inside at least three major municipal departments, according to publicly available procurement documents reviewed this week.

The timing is not accidental. Brazil's Lei Geral de Proteção de Dados, in force since 2020, requires public bodies to maintain accurate and non-redundant personal data records. Enforcement pressure from the Autoridade Nacional de Proteção de Dados has intensified since early 2026, and São Paulo, as the country's largest municipal data holder, faces the heaviest scrutiny. Duplicate image records — scanned identity documents, cadastro photos, urban planning imagery — represent both a compliance risk and a storage cost that compounds monthly.

Inside the Prefeitura de São Paulo, the Secretaria Municipal de Inovação e Tecnologia, headquartered near Viaduto do Chá in the Centro district, has been piloting a deduplication protocol since March 2026 across the city's SIGA system, the integrated administrative platform that handles everything from business licences to social assistance records. The Instituto de Pesquisas Tecnológicas, the state-linked research body based in Cidade Universitária, has been brought in to validate the algorithmic approach the secretaria is testing — specifically whether perceptual hashing, rather than simple file-size matching, should become the standard for identifying duplicate images across legacy databases.

What the Data Shows — and What It Costs

Storage is not abstract. The Prefeitura's own budget documents for 2026 allocate R$47 million to digital infrastructure maintenance across municipal systems, a figure that includes server costs at the Centro de Processamento de Dados municipal facility in Itaquera. Industry benchmarks used in comparable Latin American city procurement — Bogotá and Buenos Aires have both published deduplication tenders in the past 18 months — suggest that unresolved image duplication can inflate active storage requirements by 18 to 35 percent in large public databases. Applied conservatively to São Paulo's scale, that points to tens of millions of reais in avoidable annual expenditure.

On the private side, the stakes are different but the decisions are equally pressing. Several São Paulo-based fintechs operating out of the Faria Lima corridor use onboarding flows that collect facial images and identity document scans. Those companies must now decide whether to invest in real-time deduplication at the point of capture — more expensive upfront, typically R$2 to R$4 per verification transaction at current vendor pricing — or to run periodic batch cleanups, which carry greater LGPD exposure in the interval between sweeps. Two procurement calls published on the Diário Oficial do Estado de São Paulo in June 2026 indicate at least some operators are moving toward the real-time model.

The Decisions That Will Define the Next Phase

Three choices are converging before the end of 2026. First, the Secretaria Municipal de Inovação e Tecnologia must decide by September whether to extend its SIGA deduplication pilot citywide or to pause pending the IPT's full evaluation report, expected in August. Second, the federal government's digital identity program, Gov.br, is expanding its biometric layer into São Paulo municipal services — a rollout that, if poorly coordinated, could create a new wave of image duplication across federated systems rather than solving the existing one. Third, the São Paulo state legislature has a data governance bill — Projeto de Lei 342/2025 — sitting in committee that would set mandatory deduplication standards for state contractors. Its passage, or failure, will set the baseline for what private operators must do.

For residents, the practical stakes are most visible in the Poupatempo service centres — the busiest is at Avenida Paulista, 807 — where document processing times can stretch when back-end records carry conflicting image entries. Officials at the secretaria have not committed to a public timeline for when residents might see faster processing as a result of deduplication efforts. What is clear is that the technical and political choices made before December will lock in São Paulo's approach for years. The window to shape those choices, for both public administrators and private operators, is closing faster than most realise.