Sixteen volunteers said AI-generated “ghosts” of deceased loved ones felt comforting and often real, even when the chatbots invented details, according to Jack Manning and colleagues.
Brief surveys were the bots’ only source material, yet participants usually overlooked factual mistakes and reacted more strongly to wrong tone or style—such as missing emojis or unfamiliar nicknames.
Human researchers screened every reply before it reached participants, limiting harmful or deceptive exchanges; Manning said that kind of guardrail is largely absent from commercial griefbot services.
Participants still flagged risks, including emotional dependence and personality changes from repeated use, but the study suggests carefully designed systems could help some people seek closure in grief.
If AI ghosts invent facts to feel real, are we sacrificing authentic memories for a comforting illusion?
As companies sell AI versions of our dead, who protects the deceased from digital puppetry and exploitation?
AI Ghosts for Grief Support: The Rise, Risks, and Regulation of Digital Afterlife Technologies in 2026
Overview
AI chatbots, known as griefbots or generative ghosts, are now being offered by tech companies to help people cope with the loss of loved ones. These systems use personal data like social media posts and voice recordings to imitate conversations, allowing users to interact with digital versions of the deceased. This technology promises a sense of continued connection, but its rapid growth raises important ethical and psychological concerns. While griefbots may provide emotional comfort, experts warn they could lead to unhealthy dependency, highlighting the need for careful consideration as these tools become more common.