São Paulo's Digital Archives Face a Reckoning Over Duplicate Images: The Key Decisions Ahead

São Paulo's municipal government is sitting on a growing problem buried in its own servers. Across platforms managed by the Secretaria Municipal de Inovação e Tecnologia, or SMIT, thousands of duplicate images have accumulated inside public-facing digital archives — from the city's official portal at prefeitura.sp.gov.br to the cultural databases maintained by the São Paulo Museum of Art, known as MASP, on Paulista Avenue. The redundancy is not cosmetic. It is costing storage budget, slowing load times on citizen-facing services and, in at least some documented cases, causing version-control failures where an outdated image replaces the correct one in official documents.

The timing is awkward. Mayor Ricardo Nunes is pushing a broader digital modernisation agenda ahead of the 2026 municipal budget cycle, and the image-duplication issue has surfaced as a concrete test case for how seriously City Hall will invest in data governance infrastructure. With federal resources under the Lula administration's Programa de Aceleração do Crescimento digital component flowing toward state and municipal tech upgrades, São Paulo cannot afford to be seen as a poor steward of its own systems.

What the Problem Actually Looks Like on the Ground

The duplication crisis is not abstract. At the Centro Cultural São Paulo, on Rua Vergueiro in Liberdade, archivists managing the institution's digital collection have dealt with repeated instances of the same photographic records appearing under different file names and metadata tags — a symptom of departments uploading assets independently without a shared image management protocol. The problem is mirrored at the Arquivo Histórico Municipal, housed near the Praça da Sé, where digitisation drives conducted between 2019 and 2024 produced overlapping image sets that have not been reconciled into a single authoritative catalogue.

Tech specialists working in São Paulo's unicorn ecosystem, particularly those around the Cubo Itaú hub in Itaim Bibi, have noted that the city's public sector lags roughly three years behind private-sector image-management practices. Organisations like Cubo-affiliated startups routinely deploy perceptual hashing algorithms — tools that can identify visually identical images even when file names differ — but municipal procurement cycles have not yet brought these tools into government contracts. A standard enterprise-grade duplicate-detection licence for a mid-sized institutional archive runs between R$18,000 and R$45,000 per year, a range well within reach of SMIT's annual technology allocation but one that requires a specific procurement decision to unlock.

The Decisions That Will Define the Outcome

Three choices now sit in front of city administrators, and each carries consequences that extend beyond storage efficiency. First, SMIT must decide by September 2026 whether to run a centralised deduplication process across all municipal platforms at once or to pilot the cleanup on a single high-traffic system — the prefeitura portal draws an average of 1.2 million unique visits per month, according to figures the city published in its 2025 transparency report, making it the highest-stakes candidate but also the riskiest place to start.

Second, officials must determine who has the authority to designate a canonical image when two near-identical versions exist. This is partly a legal question. Images in the Arquivo Histórico Municipal carry heritage classifications under federal law, meaning that replacing even a duplicate requires sign-off that crosses municipal and federal jurisdiction. Getting that process wrong could trigger challenges under the Lei de Acesso à Informação, Brazil's Freedom of Information framework enacted in 2011.

Third, and most consequentially, the city must decide whether to build a permanent image-governance protocol into its next multi-year technology contract — currently set to be tendered in the first quarter of 2027 — or treat the current cleanup as a one-off exercise. Without a standing protocol, the same duplication will regenerate within eighteen months, according to the standard trajectory observed in comparable municipal digitisation projects in Bogotá and Mexico City.

Advocacy groups focused on open data, including those that regularly mobilise along Paulista Avenue during transparency campaigns, have begun pressing SMIT to publish its deduplication methodology publicly once it is chosen. The argument is straightforward: if public images are being removed or consolidated, citizens have a right to understand the criteria. The next SMIT public consultation on digital services is scheduled for August 14 at the Centro de Formação Cultural Cidade Tiradentes — and image governance is expected to be on the agenda.