Fundación Vía Libre and civil coalition challenge to Buenos Aires Fugitive Facial Recognition System (SRFP) (2019–ongoing)

In April 2019 the City of Buenos Aires launched the Sistema de Reconocimiento Facial de Prófugos (SRFP — Fugitive Facial Recognition System), a network of approximately 300 cameras installed in the Buenos Aires subway and cross-referenced in near-real-time against the CONARC (Consulta Nacional de Rebeldías y Capturas) national fugitive database, using facial recognition software supplied by local company Danaide S.A. and built on technology developed by Russia-based NtechLab. Human Rights Watch wrote to Mayor Horacio Rodríguez Larreta in October 2020 demanding immediate suspension and raising documented concerns about children's rights and algorithmic accuracy; in December 2020 the Observatorio de Derecho Informático Argentino (ODIA) filed a constitutional amparo action against the city government. Fundación Vía Libre — the Córdoba-based Argentine civil-society organisation that is the corpus's principal Argentine anchor on AI-and-human-rights work — joined the case as amicus curiae alongside the Centro de Estudios Legales y Sociales (CELS) and the Asociación por los Derechos Civiles (ADC), contributing technical and human-rights arguments about the system's legal basis and the absence of safeguards. The legal challenge produced a suspension order in April 2022, a first-instance declaration of unconstitutionality in September 2022, and an appellate confirmation of that ruling in April 2023 — the highest-order domestic judicial declaration against biometric mass surveillance in public transport yet issued in Latin America. As of early 2024 the system remained suspended while the civil coalition and city government remained deadlocked over the conditions and methodology for an independent audit required before any reinstatement.

The SRFP: system architecture and legal authorisation

The SRFP was introduced with a mandate to identify individuals whose faces matched entries in CONARC — the national register of outstanding arrest warrants — using live feeds from subway surveillance cameras. The NtechLab engine, which underlies the system's identification capability, was chosen by Danaide S.A. through a private tender process that did not follow the transparency requirements ordinarily applicable to public procurement of state surveillance infrastructure. The Buenos Aires city government authorised the system through Resolution 398 and subsequently incorporated it into local public security law via Law 6.339, passed in October 2020 — the latter a retroactive legislative act that authorised a system already operational since 2019, creating a statutory framework for technology the city had already deployed without one.

Neither Resolution 398 nor Law 6.339 required a data-protection impact assessment before the SRFP began operation. The system's operational parameters — the specific camera locations, the database cross-reference architecture, the data retention periods, the access control framework — were not disclosed publicly and the city government refused freedom of information requests for procurement documents, classifying them as confidential. The first independent public accounting of the system's actual footprint came through the judicial inspections ordered in the ODIA amparo.

HRW's October 2020 letter and the children's rights concern

Before the formal legal challenge, Human Rights Watch wrote to Mayor Rodríguez Larreta on 9 October 2020 demanding the SRFP's immediate suspension and raising two concerns the subsequent litigation would confirm at a doctrinal level. The first was a children's rights violation: at least 166 minors had been listed in CONARC between May 2017 and 2020, exposing them to identification under a system trained primarily on adult faces; NtechLab's algorithms had shown six-times higher false-match rates for children aged 10–16 compared to adults aged 24–40, and the system had been tested only on police employee faces with no bias-minimisation checks for any demographic group, including children. HRW argued this violated the Convention on the Rights of the Child — which protects the anonymity of minors in criminal proceedings — and the UN Standard Minimum Rules for the Administration of Juvenile Justice (the Beijing Rules). The second was the broader demand for a privacy and human-rights impact assessment before any continued operation, an assessment the city had not conducted and had not committed to conducting. HRW's letter was the campaign's first internationally visible advocacy act and established the algorithmic-bias-and-children's-rights framing that the appellate court would later embed in its reinstatement conditions, requiring investigation of whether the system produces "differentiated impact depending on personal features".

The ODIA amparo and Vía Libre's amicus role

In December 2020 ODIA filed an amparo — the constitutional-protection action provided by Article 43 of Argentina's Constitution — against the City of Buenos Aires Government before the administrative courts, challenging Resolution 398, Law 6.339, and the system's implementation on grounds of unconstitutionality and non-compliance with Convention 108 on automatic data processing and with Buenos Aires city public security laws 5688 and 6339. Fundación Vía Libre joined the case as amicus curiae, submitting technical AI-systems analysis and human-rights arguments about the adequacy of the legal basis, the absence of a data-protection impact assessment, and the structural risks of deploying high-error-rate biometric surveillance in public transport without any oversight architecture. CELS contributed the human-rights litigation architecture that anchors its long experience in amparo actions against the Argentine state; the Asociación por los Derechos Civiles (ADC) also participated. Judge Andrés Gallardo ordered inspections of the Urban Monitoring Centre and the Ministry of Security — the forensic examination that produced the factual record on which the unconstitutionality ruling ultimately rested.

The April 2022 suspension and factual disclosures

The court-ordered inspections made public a set of facts the City of Buenos Aires had not previously disclosed and that were central to the April 2022 suspension order. The SRFP had been used to make more than nine million biometric queries against a CONARC database of only around 35,000 to 40,000 active fugitive entries — a nine-million-to-35,000 ratio that demonstrated the system was scanning a population orders of magnitude larger than the stated fugitive-identification purpose could justify. The biometric data of approximately 7.5 million Buenos Aires residents had been accessed without warrant. Seventeen anonymous administrative accounts were found to have unrestricted access to RENAPER (Registro Nacional de las Personas) national identity data well beyond what CONARC cross-referencing required. Three hundred and fifty-six search records had been manually erased. Among those whose biometric data had been accessed without any criminal warrant were journalists, politicians across party lines, human rights leaders including Estela de Carlotto (president of the Abuelas de Plaza de Mayo), and members of Argentina's Supreme Court — a surveillance pattern whose breadth confirmed that the SRFP's actual operational scope extended well beyond fugitive identification into population-wide monitoring of public figures.

The wrongful detention that became the campaign's most publicly cited individual harm was the case of Guillermo Ibarrola, a factory worker who was arrested in Buenos Aires and held for nearly a week in Bahía Blanca — a city almost 400 miles away — before authorities discovered that a data-entry error had caused his national ID number to be registered in CONARC under the name of a wanted fugitive who shared his surname. Ibarrola's case demonstrated that the system's harm pathway ran not only through algorithmic misidentification but through upstream data-quality failures in the underlying fugitive registry — any system that produces detentions on automated biometric matches can generate injustice before the algorithm itself has a chance to fail, through the data structures the algorithm queries.

The September 2022 unconstitutionality ruling

On 7 September 2022 City Administrative Judge Elena Liberatori declared the SRFP's implementation unconstitutional, prohibiting operation of the system until the City of Buenos Aires established the legal and oversight architecture the ruling required. The first-instance judgment found that the SRFP had been deployed without due protection of the rights of Buenos Aires residents, contrary to the presumption of innocence — the ratio of nine million queries against 35,000 fugitive entries demonstrating that the system treated the entire Buenos Aires subway-riding population as suspect — and without the Special Committee for Video Surveillance Monitoring, the surveillance Registry, and the impact assessment that Buenos Aires city laws 5688 and 6339 required as conditions of lawful biometric deployment. The ruling specified four conditions for any reinstatement: establishment of a functional Special Committee; a proper surveillance Registry as required by law; a full data-protection impact assessment; and a public consultation process that gave Buenos Aires residents a meaningful opportunity to engage the necessity and proportionality of the system. Until those conditions were met, the SRFP would remain prohibited.

The April 2023 appellate confirmation

On 28 April 2023 the Court of Appeals of the City of Buenos Aires, Chamber I, confirmed the unconstitutionality ruling and upheld the suspension, rejecting appeals submitted by the City Government and the Public Prosecutor's Office. The appellate ruling confirmed the first-instance judgment's constitutional analysis and modified the conditions for reinstatement with heightened specificity: in addition to the oversight and impact-assessment requirements from the September 2022 ruling, the appellate court added a requirement that the city must investigate whether the SRFP produces differentiated impact depending on personal features — the demographic algorithmic-bias question that HRW had raised regarding children in 2020 and that the civil coalition had maintained applied to other demographic groups as well. This appellate ruling elevated the Buenos Aires case from a first-instance administrative finding to confirmed constitutional doctrine within the city's judicial hierarchy: a higher-order precedent for any future Argentine government seeking to deploy biometric mass surveillance in public infrastructure.

The audit standoff and ongoing suspension

The reinstatement conditions set by the September 2022 ruling and confirmed in April 2023 require an agreement between the city government and the civil society groups on an institutional framework, budget, and methodological plan for a system audit before the SRFP can reactivate. In February 2024 the court found the parties deadlocked. The city government had proposed a form of "black box" audit — operational testing of the system's performance without independent access to the system architecture, training data, or algorithmic specifications — which the civil coalition, led by Fundación Vía Libre's Beatriz Busaniche and the ODIA and CELS teams, rejected as structurally inadequate for critical public-surveillance infrastructure. Busaniche framed the coalition's position: "facial recognition is still too unreliable to be deployed freely", and the unresolved question of "who does the audit and how" could not be resolved by a test the operator itself designed. The city continued to hold public seminars on the system as part of a reintroduction preparation process. The civil coalition continued to oppose reinstatement on those terms. As of early 2024, the system remained suspended with no reactivation date.

Significance for the broader AI-good movement

The Buenos Aires SRFP campaign is the corpus's first Argentina-anchored civil society campaign on biometric mass surveillance and the first entry in the Southern Cone's domestic legal-challenge register on facial recognition in public transport infrastructure. Its significance for the wider make-AI-good movement runs on three connected axes.

First, the judicial outcome represents the highest-order domestic judicial declaration against deployed biometric mass surveillance in public transport yet entered in Latin America. The September 2022 first-instance ruling and its April 2023 appellate confirmation — together with the appellate court's affirmative requirement of demographic-bias investigation before reinstatement — placed the Buenos Aires case inside the small global cohort of judicial decisions holding that a biometric surveillance system deployed at scale without a data-protection impact assessment or oversight mechanism is constitutionally defective by design rather than merely administratively irregular. The ruling's core structural reasoning — that the nine-million-to-35,000-entry query ratio demonstrated the system's actual purpose extended well beyond fugitive identification into population-wide surveillance — is the case's most transferable doctrinal contribution to the Latin American and broader civil society field challenging the "law enforcement limited purpose" framing under which face-recognition surveillance is typically licensed.

Second, the campaign is the corpus's most fully documented instance of the ODIA-CELS-Vía Libre Argentine civil society coalition operating as a constitutional-litigation machine on AI-and-human-rights questions: Fundación Vía Libre supplying the technical AI-systems analysis and policy argument, CELS the human-rights litigation architecture from its deep experience in amparo actions against the Argentine state, and ODIA the computer-law and digital-rights doctrinal frame. The coalition subsequently appeared together at the March 2025 Inter-American Commission on Human Rights hearing on AI and human rights — a continuity that confirms the campaign has functioned as an organisational alignment-building event for the Argentine civil AI-accountability field rather than a single-event litigation. That continuity connects the Buenos Aires case to the Derechos Digitales IACHR hearing in the corpus's Latin American coverage of multilateral AI accountability mechanisms.

Third, the children's rights angle — documented through the NtechLab algorithm testing data HRW cited in 2020 and subsequently embedded in the appellate court's demographic-disparate-impact condition — is the corpus's first Latin American entry in which documented differential algorithmic harm to a specific demographic produced both an international advocacy letter and a binding judicial condition before reinstatement. The case connects the corpus's facial-recognition resistance thread — anchored by Coding Rights in Brazil, Internet Freedom Foundation in India, and EDRi's Reclaim Your Face coalition in Europe — to the Argentine civil society field at the level of domestic constitutional doctrine, and establishes the Buenos Aires case as the regional reference point for amparo as a vehicle for securing judicial review of deployed biometric surveillance systems in Latin American jurisdictions.