AI Grief Bots

AI grief bots are post-mortem avatars or conversational simulations of people who have died. In The ethics of using AI to immortalize the dead, Tomas Holoneck explains that they may be built from materials a person prepared before death or from personal data uploaded later by surviving loved ones.

The concept matters because it turns ordinary Persistent Agent Memory and AI Companion Active Memory into a post-mortem representation problem. A useful assistant memory can remember preferences or conversations; a grief bot may be perceived as carrying a person’s voice, style, private relationships, or legacy after that person can no longer approve, object, clarify, or withdraw.

Key Claims

• Grief bots are built from selected personal traces, such as messages, emails, videos, and other records.
• The ethical risk depends on who selected the data, what the deceased person consented to, and who can later access or commercialize the representation.
• A post-mortem avatar can comfort one survivor while unsettling or harming another, so family context and individual consent matter.
• General assistants such as ChatGPT can be used for this kind of simulation, making the risk broader than dedicated memorial products.
• The category overlaps with Digital Memorialization, but interactive simulation is more ethically charged than static archiving.
• Public institutions and NGOs may be needed because profit-seeking companies have incentives to monetize grief and dependency.

Connections

• Tomas Holoneck, University of Cambridge, Marketplace Tech, and Stephanie Hughes - source context.
• Post-Mortem AI Consent - consent, dignity, and privacy boundary.
• Digital Memorialization and Digital Preservation - broader memory and archive branch.
• Persistent Agent Memory, AI Companion Active Memory, and Human Connection Under AI - adjacent AI memory and relationship concepts.
• ChatGPT - general assistant surface that can be used for conversation-like simulation.