The Hidden Cost of Talking to the Dead Online

When Marina Smith died in 2022, she attended her own funeral. Smith, a Holocaust educator, had spent months recording her life story with a company called StoryFile. The technology didn’t resurrect her — it selected from hours of pre-recorded answers, using AI to match mourners’ questions to something she had actually said. For her family, the result was an uncanny semblance of presence: a face on a screen that spoke in her voice, as if she were still in the room. Tech coverage framed the experience in terms of comfort. StoryFile described its product as a way to “say goodbye in a new way,” and a wave of grief-tech startups has built on that promise. Somnium Space, a virtual-reality platform, is developing a “Live Forever” mode — inspired by the founder’s loss of his father — that will capture full-body movement and biometric data through a haptic Teslasuit, eventually letting a person’s digital likeness converse with the bereaved long after the biological body is gone. The dominant narrative is that these tools are therapeutic: a gentle bridge across the void, a way to keep a bond alive. Clinicians I follow are starting to push back — not against the desire to remember, but against the architecture that makes it hard to let go. Therapists have begun to describe a pattern they call “grief entrapment”: when the technology designed to comfort becomes a mechanism that stalls the natural work of mourning, keeping the bereaved in a loop of near-constant interaction with a representation that never changes. Normal grieving, the evidence from bereavement research tells us, involves a gradual restructuring of the relationship to the deceased — not cutting ties entirely, but retuning them so they no longer demand the same kind of ongoing response. A bot that can be summoned at any moment, that replies with plausible warmth, works against that gradual disengagement. The person is not present, but the signal keeps coming. In media culture terms, what we’re seeing is the collision of two older tensions. One is the deep attachment people form with digital selves, an attachment that makes the loss of an avatar or a platform feel like a genuine bereavement. The other is the drive to encode personality and memory into durable media — a drive that has produced everything from the Tupac hologram at Coachella in 2012 to the current wave of “digital twin” experiments. Combine them, and you get a grief economy that sells continuity without asking what kind of continuity serves the mourner. The psychological reality of avatar attachment is well documented. A 2025 study in Frontiers in Psychology that examined virtual mourning in VRChat found notably different ritual styles: younger users often processed loss through humor and narrative play, while older participants constructed more solemn memorials, seeking symbolic continuity. That’s a reminder that the need to mark a digital death is real, and that it emerges in culturally shaped forms. The problem with the immortality bot is not that it’s digital — it’s that it removes the endpoint those rituals are supposed to navigate toward. If you can talk to the dead person every night, the stages of mourning lose their temporal scaffold; the sense of finality that eventually allows the living to reinvest in new relationships becomes elusive. And yet, other corners of the metaverse are experimenting with the opposite approach. EvolVR, a VR grief support group covered by MIT Technology Review, uses the disinhibition of the digital veil to let people express vulnerability they might suppress in a physical room. Here, the avatar is not a stand-in for the deceased but a container for the living person’s pain — a way to be fully present with loss, not a way to outrun it. Some communities even practice deliberate detachment: virtual meditation chambers where long-term metaverse residents train themselves to imagine their digital lives erased, training a kind of Stoic muscle so that losing a platform or a persona doesn’t shatter them. That’s not apathy; it’s a cultivated capacity to distinguish between the self and its projected extensions. The contrast should give designers pause. The tools we build to mediate grief can either help people process the reality of absence or offer them a seductive escape from that reality. The difference lies in the framing, the temporal boundaries, and the intent baked into the interface. A chatbot that promises unlimited interaction, marketed as “never lose someone again,” leans hard toward entrapment. A structured experience that offers a final conversation — perhaps to hear a message the deceased left specifically for a spouse, or to speak words the mourner couldn’t say before — might serve a genuine palliative function without overriding the human need for resolution. What is ethically required is not a blanket rejection of grief tech. The integration of digital presence into mourning rituals is not fundamentally different from keeping a photograph or a recording; it’s an intensification, not a category change. But intensification changes the stakes. When the design defaults to permanence, it is worth asking whether we are building for healing or for something more like a comfortable trap. Good palliative design would embed clear limits — a finite number of interactions, a session that deliberately ends — and would frame the experience as part of saying goodbye, not indefinitely postponing it. The most valuable thing a grief technology can offer is not the illusion that death didn’t happen. It’s the scaffolding that lets the living acknowledge what has been lost, honor what remains in memory, and step, eventually, into a world that no longer includes that voice — except in the irreplaceable privacy of the mind. That’s not a cold limitation. It’s what mourning has always demanded, and what any tool that claims to serve the bereaved has to respect.