Narrative review uncovers how generative artificial intelligence alters mourning practices in contemporary society, highlighting the need for clinical frameworks and privacy regulations.
Background: The accumulation of digital traces and generative AI has transformed death from a private biological event into a persistent, interactive social presence. This challenges detachment models of grief and revitalizes continuing bonds theory, while creating new vulnerabilities including fear of second loss, algorithmic grief triggers and commercialization of mourning. Methods: We conducted a narrative review following SANRA guidelines. Literature was searched across Scopus, Web of Science, PsycINFO, ACM Digital Library and Google Scholar (2010-2026, focus 2022-2026) for terms related to digital legacy, digital afterlife, online mourning, continuing bonds, griefbots/deadbots, generative ghosts and post-mortem privacy. Thirty-six influential peer-reviewed articles, monographs and conference papers were retained for thematic synthesis. Findings: We identified three interlocking transformations: from passive archiving to active simulation where griefbots generate novel utterances, from private time-limited to public, networked and algorithmically curated mourning with temporal, spatial and social expansion and platform vernaculars like GriefTok, and from peripheral to central ethics. Building on Price et al. (2025), we propose a four-type typology of digital continuing bonds: passive, active one-way, interactive, and communal. Adaptive value depends on internalized versus externalized bonds, perceived control and offline support. Generative ghosts create ambiguous presence that can scaffold coping through affective scaffolding and habits of intimacy, but also disrupt lived temporality and distort collective memory. Interpretation: Results inform clinical assessment using Dual Process Model, responsible design principles (opt-in consent, transparency, user control, sunset clause) and regulation of post-mortem privacy at EU level, given the projected 1.4-4.9 billion deceased profiles by 2100.
No takes yet. Share an insight, caveat, or question.
Okoń et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: