Navigating Sensitivity, Scarcity, and Scale: AI and the Future of Sensitive Data Curation in Small Social Science Archives

The curation of sensitive social science data is becoming increasingly complex in the face of rapid technological change, shrinking resources and stricter ethical and legal requirements. Small domain-specific data archives such as the Slovenian Social Science Data Archives (ADP) are under pressure as they have to balance long-term data stewardship with the growing demands for rapid and accessible data publication. While quantitative data is often anonymised by researchers, qualitative material such as interviews requires expert review to identify identifiers — an effort that requires a lot of time and resources. AI offers potential relief by automating tasks such as entity recognition and redaction. However, its use raises significant concerns about the legal and ethical implications of machine processing, particularly in the sensitive pre-publication phase when archivists prepare records for secure transfer. For archives with limited technical capacity, the independent implementation of such tools is a challenge. This lightning talk highlights ADP’s role in CESSDA’s collective efforts to explore AI in digital curation and develop collaborative solutions for small archives with ethical oversight. It advocates inclusive, federated AI approaches that ensure data integrity, privacy and trust.