Archiving Queerly Otherwise: Queer Records, AI, and the Potentialities of Continued Harm

This paper examines the ethical tensions of artificial intelligence (AI) technology use in queer archival contexts. As AI tools are increasingly adopted in information workflows, (e.g. transcription, discovery), there is a growing need to critically assess how such technologies interact with records documenting LGBTQ + lives. The transformation of queer histories into machine-readable formats raises concerns about visibility, misclassification and the potential for continued and renewed forms of archival harm. Drawing on critical archival and classification studies, queer theory, digital humanities, and the Australian context, this paper argues that these technologies must be evaluated in relation to the values that guide archival and library practice, and potential harms. Through illustrative vignettes, risks such as posthumous outing, epistemic flattening and the unintended consequences of data-driven classification are highlighted. These scenarios speak to broader questions facing the archival, library, and information professions of emerging technology applications upholding equity, justice, privacy and human dignity. Rather than positioning AI as an inevitable innovation or dismissing it entirely, proposed is a cautious, community-centred approach through an ethical pre-condition framework. This framework aims to foreground ethical responsibility and stewardship, and recognises the right (not) to be classified.