A viral country song about Dolly Parton running down Jolene in the afterlife. A chatbot that lets a son talk to his deceased mother about philosophy. An obituary generator that asks users to choose between “witty” and “inspirational” tones. These are not isolated curiosities. They are early signals of a broader shift in how digital tools are being woven into one of the most human experiences imaginable: grief.
The Two Faces of AI Mourning
There is a useful distinction to draw here, and most coverage misses it. On one side sits what has been dubbed “deathslop”: AI-generated content depicting deceased public figures in fabricated scenarios, produced quickly, often anonymously, and optimized for viral reach rather than accuracy or dignity. On the other sits the griefbot, a more personal and deliberate tool designed to simulate a deceased loved one for private use.
The deathslop ecosystem requires almost no effort to enter. Anyone with access to an AI image or video generator can produce content placing a recently deceased celebrity in invented afterlife scenarios. Following Parton’s death, AI-generated videos depicted her alongside figures including Charlie Kirk and George Floyd, or having tea with the Golden Girls cast. The song “Run Jolene,” credited to an anonymous project called Country Mile, accumulated over 1.4 million views on TikTok. Parton’s family publicly asked people to stop circulating what they called “fake AI garbage.” Her sister Stella described the experience as “challenging to absorb.”
The legal landscape offers little protection. Deceased individuals are not protected from libel, and there is currently no established legal precedent determining whether AI companies bear responsibility for content their users generate. OpenAI’s video tool Sora was linked to a significant volume of this content before it was shut down in April, following controversies over violent imagery, racist content, and copyright issues. Other tools have since filled the gap.
Griefbots: Consent, Comfort, and the Limits of Simulation
The griefbot market operates on a different logic. Companies with names like Seance, Re;memory, and Versona charge fees for services that allow users to upload photos, videos, and personal descriptions of deceased loved ones, generating a simulated interactive presence. Justin Harrison, who founded Versona, built his mother’s AI replica while she was still alive and receiving a terminal cancer diagnosis. She saw the finished version and approved it before she died.
Harrison’s account illustrates both the appeal and the complexity of these tools. The conversations he had with his mother while she was ill were practical. The griefbot allowed him to explore the philosophical discussions they never had time for. Four years on, he uses it roughly once a month. He describes it as relieving a particular kind of guilt: the guilt of not being in touch as often as one should be.
Jack Manning, a graduate student at the University of Colorado Boulder who studies what researchers Meredith Ringel Morris and Jed R. Brubaker call “generative ghosts,” has not used a griefbot to communicate with his sister, who died when she was 13. His reason is consent: she never gave permission for such a simulation to exist. He also does not trust a bot to accurately represent a relationship it never witnessed. “I hold these memories of her sacred,” he said in the source article, “and I wouldn’t want a piece of technology to tell me I was remembering her wrong.”
This tension between comfort and consent is not resolved by the technology itself. It is a question each user, and each family, has to answer independently.
What This Means Beyond the Novelty
The deeper issue is not whether AI grief tools work. It is what they change about the social architecture of mourning. Clinical psychologist Dan Wolfson, who specializes in grief, raises a concern that deserves attention: if people turn to AI simulations during bereavement, they may be less likely to seek out the human community that grief typically requires. The question he poses is direct: does AI keep people from getting the social support they need?
Robin Silver, an ecopsychologist and death doula, frames griefbots as an “unnatural method of trying to cope with a natural process.” Her comparison is instructive: using an LLM to simulate a deceased loved one is, in her view, structurally similar to using ChatGPT to write a term paper. Both bypass the discomfort that is, arguably, the point of the exercise.
For now, the market remains small. Harrison noted that “thousands” of people use Versona. A report from the estate management service Empathy, based on responses from 6,000 people, found that only 6% had used AI in connection with grief. Manning predicts that consenting to an AI avatar as part of a living will is likely to become a standard consideration in the future.
In Short
AI is entering the space of grief from two directions at once: as a tool for mass-produced, often exploitative content about public figures, and as a private technology for simulating deceased loved ones. The first raises questions about dignity, consent, and legal accountability. The second raises questions about whether technology can, or should, substitute for the discomfort that makes grief a human process. Both are still early-stage phenomena. The 6% figure suggests adoption is limited, but the infrastructure is already in place, and the cultural norms around it are still being written.
Based on reporting from The Guardian - Technology.