Beatriz Busaniche

Beatriz Busaniche is the Buenos Aires-based activist, researcher, and academic serving as president of Fundación Vía Libre — the Córdoba-headquartered Argentine civil-society organisation that anchors the Spanish-language Latin American work on the social impacts of artificial intelligence, electronic voting, privacy, surveillance, free software, and access to knowledge (see Person entry for affiliations, academic posts, and biographical structure). She is tracked here as a Voice because her two decades at Fundación Vía Libre have produced a sustained public analytical register — running from the early 2000s free-software and copyright-reform arguments through the sustained electronic-voting opposition campaign and into the current AI-and-human-rights line centred on the EDIA participatory bias-audit toolkit and HESEIA community-based bias evaluation dataset — whose most load-bearing single formulation is the structural claim that democratic accountability of digital infrastructure must be established before the AI conversation proceeds. She is the corpus's first Argentine Voice entry, closing the Argentina gap in the Latin American voice cluster alongside the already-anchored Paraguayan (Maricarmen Sequera), Chilean and regional (Jamila Venturini, Juan Carlos Lara) and Colombian (Carolina Botero) Latin American voice anchors.

"Free software, free culture, free knowledge": the foundational arc

The public register Busaniche has carried for over two decades begins not with AI but with the question of who controls the infrastructure of digital knowledge. She self-describes as "an activist for Free Software, free culture and human rights in environments mediated by ICTs" — a formulation that names the three-register arc her work occupies. The free-software register grounds her in the Argentine free-software movement's early-2000s institutional moment. The free-culture and access-to-knowledge registers extend the argument from software licensing into cultural production and knowledge commons more broadly: she co-founded Wikimedia Argentina in 2007 and has been a long-standing member of the Creative Commons Argentina team — building the open-knowledge infrastructure that makes the free-knowledge argument operational rather than rhetorical.

Her authorship of Monopolios Artificiales sobre Bienes Intangibles (2007) and Propiedad Intelectual y Derechos Humanos (2016) gives the register its textbook dimension: these are the teaching texts — used in her courses at the Universidad de Buenos Aires, Universidad Torcuato Di Tella, and FLACSO — through which she has carried the intellectual-property-and-human-rights argument into Argentine academic discourse across two decades. The structural argument the texts develop — that artificial monopolies on intangible goods (software, cultural works, information) represent a form of rights restriction with a coherent human-rights analysis — is the foundational move on which her current AI-and-human-rights work rests: before the AI conversation, the digital-infrastructure conversation; before the digital-infrastructure conversation, the access-to-knowledge conversation. Each layer is load-bearing for the next.

Electronic voting: algorithmic accountability as precursor

The longest-running specific campaign in Busaniche's public output — predating the AI-and-human-rights register by a decade — is Fundación Vía Libre's sustained opposition to electronic voting in Argentine federal and provincial elections. Listed under the programme area "Civil and Political Rights and Electoral Technologies," the campaign has targeted the Boleta Única Electrónica (BUE) electronic voting machines introduced in Buenos Aires City elections from 2015 onward and the subsequent proposals to extend the system nationally. The security-and-auditability argument the campaign advances is that electronic voting systems cannot provide the same public verifiability guarantees as paper-based systems: the BUE code is proprietary, external audit access is restricted, and the vote count is therefore structurally non-transparent to ordinary citizens in a way that paper-ballot systems are not. The 2025 year-in-review records Ekoparty's lifetime-achievement recognition of Vía Libre's electronic-voting security audit work — at a security and hacker conference — as evidence that the technical-community standing of this campaign is independently established, not solely a civil-society advocacy claim.

The campaign is distinctive in the corpus because it is pre-AI algorithmic accountability work: the argument that a consequential decision-making system must be auditable by ordinary citizens, not just by technical specialists with insider access, is precisely the structural argument that the EDIA bias-audit toolkit extends a decade later into AI. The democratic-accountability demand — public source code, verifiable outputs, no proprietary black boxes in democratic infrastructure — is the same demand whether the infrastructure in question is electronic voting or large language models in state services. The arc is linear.

"Hate speech is sustained by stereotypes, which are often driven by AI"

The most concentrated single formulation in Busaniche's public AI-and-human-rights output is her canonical framing, delivered to the Open Knowledge Foundation in April 2023: "hate speech is sustained by stereotypes, which are often driven by AI." The framing collapses three analytical moves into a single sentence — that AI systems encode and amplify existing social stereotypes; that stereotypes are the substrate on which hate speech propagates; and that AI governance therefore cannot be separated from anti-discrimination policy. It is the thesis that organises Fundación Vía Libre's AI programme and, more specifically, the EDIA project: if AI systems driven by stereotypes sustain hate speech, the intervention point is stereotype identification and audit by the communities that bear the discriminatory costs, not model audit by technical specialists who lack the contextual knowledge to identify what counts as a harmful stereotype for a given community.

The EDIA toolkit — Estereotipos y Discriminación en Inteligencias Artificiales — is Busaniche's most load-bearing single public-output artefact in the AI register. Co-developed alongside Laura Alonso Alemany and Luciana Benotti of the Universidad Nacional de Córdoba's FAMAF faculty, EDIA is an open-source tool designed to let non-technical users with lived experience in discrimination explore, characterise, and audit biases and stereotypes in word embeddings and large language models — supporting English and Spanish base models out of the box. The participatory design is the political argument made technical: datasets are built by communities that live under the stereotypes being audited, not by NLP researchers supplying externally-derived "bias" taxonomies. The project was selected in 2024 as one of five Mozilla Data Futures Lab Infrastructure Fund awardees and in 2025 won the Mozilla Festival audience-vote impact award in Barcelona. The academic record of the project runs through the 2023 ACL workshop paper Bias assessment for experts in discrimination, not in computer science and the 2024 Communications of the ACM article Exploring Stereotypes and Biases in Language Technologies in Latin America — evidence that Vía Libre's hybrid civil-society / academic position has made the EDIA project a productive NLP-literature intervention, not only a civil-society advocacy tool.

HESEIA: the classroom as community dataset

The second anchor in Busaniche's AI register is HESEIA, described on Vía Libre's own home page as "a community-based dataset for evaluating social biases in large language models, co-designed in real school settings in Latin America." HESEIA applies the EDIA participatory principle to the dataset-construction problem at scale: rather than having researchers decide what examples of stereotyping to include in a bias-evaluation dataset, the dataset is built by students and teachers in Argentine and Latin American school settings, supplying the situated cultural knowledge that determines what counts as a harmful or distorted representation in their specific linguistic and social context.

In June 2024 Vía Libre and FAMAF launched the extension course Tools to Explore Biases and Stereotypes of Artificial Intelligence in the Classroom, which attracted more than 500 registrants across Córdoba province — 34.1% from outside the provincial capital and 55.6% with no prior AI-tool experience. The reach figure is the political argument made operational: the communities best placed to identify AI-encoded stereotypes are the ones who live under them, and reaching those communities at scale requires building into educational infrastructure rather than into specialist civil-society circuits. The HESEIA / EDIA combination was accepted at EMNLP 2025 and a related paper recognised at ICLR 2026 — placing a community-based, classroom-co-designed Latin American bias evaluation dataset in the same peer-reviewed venues that the NLP field's mainstream benchmark datasets occupy.

The IACHR hearing and Argentine AI governance

The most recent layer of Busaniche's public voice is a direct AI-governance intervention at the regional human-rights mechanism level. On 7 March 2025 she appeared at the Inter-American Commission on Human Rights' regional hearing on Artificial Intelligence and Human Rights on behalf of a four-organisation Argentine civil-society coalition — Centro de Estudios Legales y Sociales (CELS), Democracia en Red, Fundación Vía Libre, and the Observatorio de Derecho Informático Argentino (ODIA) — raising concerns about the Argentine state's adoption of predictive-algorithm surveillance technologies in the absence of adequate legislation, accountability mechanisms, or public deliberation. The hearing sat alongside a wider regional civil-society submission co-signed by seventeen Latin American organisations; Vía Libre's contribution centred on the Argentine sub-case, including the multi-year challenge to the City of Buenos Aires's Fugitive Facial Recognition System (SRFP) — in which ODIA's amparo lawsuit (joined by Vía Libre as amicus curiae) had produced the September 2022 first-instance unconstitutionality ruling and the April 2023 Court of Appeals confirmation, among the most consequential judicial precedents on public-space algorithmic surveillance in Latin American constitutional law.

The IACHR hearing demonstrates the institutional depth of Busaniche's voice: from a participatory classroom dataset-building programme in Córdoba schools at one end of the spectrum, through a civil-society interlocutor role at the principal regional human-rights mechanism at the other, with academic NLP-literature publications bridging the two. The distinctive shape of a voice that has, over two decades, consistently refused to separate the technical argument (this system does not work as claimed) from the democratic accountability argument (this system must be publicly auditable by the communities it affects before it can be deployed).

Why this is a Voice entry

A Voice entry is created here, rather than additional structure on the Person entry, because Busaniche's public analytical register is itself the load-bearing object the corpus needs to track: the Argentine argument that AI governance requires democratic infrastructure accountability first — that the free-software, copyright-reform, and electronic-voting-accountability campaigns of the 2000s and 2010s are preconditions for the AI-and-human-rights work of the 2020s, not separate lines — carried through the EDIA participatory bias-audit toolkit (Mozilla Data Futures Lab 2024; Mozilla Festival 2025 impact award; Communications of the ACM 2024; EMNLP 2025), the HESEIA community classroom dataset, and the IACHR regional AI governance hearing. The corpus carries no Argentine Voice before this entry; this closes that gap. The nearest Latin American Voice anchors — voice-jamila-venturini.md (Brazil, Derechos Digitales regional AI and data colonialism), voice-juan-carlos-lara.md (Chile, Derechos Digitales co-direction), voice-carolina-botero.md (Colombia), voice-maricarmen-sequera.md (Paraguay, TEDIC) — all carry country-and-region-specific registers; this entry adds the Argentine national-jurisdiction register with its distinctive long-arc free-software-to-AI trajectory and the civil-society / academic EDIA-HESEIA hybrid that occupies neither the pure advocacy nor pure research slot. Affiliation, academic posts, recognitions, and biographical detail are recorded on the linked Person entry per the corpus's Person/Voice split.