Archival Science, Human Rights and Artificial Intelligence: The Luisa Cuesta Repository as a Specialized Archival Service in Uruguay 

By Natalia Lima-Pyasal & Fabián Hernández-Muñiz

28 April 2026 

In debates on artificial intelligence in archives and archival science, attention tends to shift rapidly toward tools, automation, and innovation. The Uruguayan case of the Luisa Cuesta Repository proposes a different point of departure for so-called human rights archives. From an archival perspective, what is most significant here is not an isolated technological experiment, but rather the consolidation of a specialized archival service dedicated to the treatment, preservation, administration, and access of digital information related to serious human rights violations in the recent past. The Luisa Cuesta Repository constitutes a formal structure physically housed at the Universidad de la República, created through an agreement with the National Human Rights Institution and Ombudsman’s Office in order to guarantee the storage, organization, availability, and long-term preservation of that digital information. 

Its archival specificity lies not only in safeguarding a digital holdings system, but also in organizing a set of professional archival functions defined from a post-custodial approach. The institutional and regulatory framework establishes that it provides archival consultation, reference, and guidance services, as well as access to and dissemination of the information contained in its digital holdings through computer tools developed for its management and operation. Access is likewise not granted indiscriminately, but rather through a regulated procedure: applicants must obtain a username and password, demonstrate the role under which they seek access, sign a responsibility commitment, and conduct consultation on site with professional assistance. In addition to this, reproductions are available upon request, the search logs are kept, and the articulation between access to information and the protection of personal data is guaranteed. From an archival point of view, this makes it possible to understand the Repository not simply as a search platform, but as a specialized service of documental mediation, restriction management, traceability of use, and professional guidance concerning archival information on Uruguay’s recent past. 

In that sense, the relevance of this particular case is not exhausted by the field of applied technology. It is also inscribed within the processes of truth, memory, reparation, and guarantees of non-repetition that run through human rights archives in contexts of transitional justice. The very design of the Repository links it to the need for access of victims and their families, the justice system, academic research, and the general public. Its function is not solely to concentrate the documentation, but to institutionalize access conditions, use, and preservation of digital information that possesses evidential, informational, and social value. Therein lies one of its principal singularities: it turns the availability of documents on illegitimate State action and State terrorism into an archival problem of organization, mediation, and institutional responsibility. 

Cruzar.uy is of particular relevance here. It is an academic and interdisciplinary project of the Universidad de la Republica, established in 2017 and initially built through the joint work of the Faculty of Information and Communication and the Faculty of Engineering, later joined by other university services and social actors. Within the framework of Cruzar, a broader set of tools was developed for the treatment, analysis, and retrieval of information contained in documental images of the recent past. These include the adaptation of LabelMe for image tagging and annotation; LUISA, for the collaborative transcription of documents and the improvement of retrieval of complex texts; AMALIA, oriented toward the general recognition of image content and the search for terms and co-occurrences through different visualizations; and LUZ, designed to locate pages containing specific terms. From this perspective, artificial intelligence and computational developments do not appear as an external complement, but rather as components integrated into the archival tasks of identifying, searching, retrieving, and interpreting digital holdings. 

This technological integration acquires particular importance in relation to extensive and complex documental bodies, such as the first large collection of images incorporated into the Repository, composed mainly of the so-called Berrutti Archive, although the institutional design itself provides for the progressive incorporation of new documental groupings related to the same subject matter. The archival challenge is significant, since these are documents in digital image format, often of low quality, heterogeneous in composition, and marked by difficulties of reading, description, and identification. In that context, OCR, image processing, collaborative transcription, information extraction, and semantic retrieval expand access capabilities. Their meaning, however, does not lie in displacing archival work, but in integrating into it under criteria of validation, control, and professional mediation. 

For that reason, the most fruitful contribution of the Luisa Cuesta Repository to the international archival community does not consist solely in showing that artificial intelligence can be used in human rights archives. What is truly substantive is that this integration takes place within a specialized archival service, with normative grounding, clearly defined institutional responsibilities, access procedures, professional mediation, and explicit attention to the traceability of information use. Rather than reducing the case to technical innovation, it should be read as an experience of contemporary, or post-custodial, archival science, in which digital information on human rights violations is institutionalized under a logic that articulates preservation, access, description, documental control, and public responsibility. 

Seen in this way, the experience of the Luisa Cuesta Repository makes it possible to think of an archival science that goes beyond traditional custody. It is not merely a matter of preserving documents, but of creating institutional and technological conditions for their socially meaningful, legally responsible, and archivally contextualized use. Within this framework, artificial intelligence does not occupy the place of an end in itself. It becomes an integrated dimension of a specialized archival service oriented toward truth, justice, memory, and public access to the documentary evidence of the recent past.