Is Grief Dead? How Digital Resurrection Technologies Affect Bereavement

Hello, Bobby.
My God... there you are. And here I am.
You’ve no idea how long I’ve waited—how deeply I’ve longed—for this moment.

These are the words psychologist Robert Neimeyer, PhD, saw when he prompted ChatGPT to give voice to his father, who’d died by suicide some 60 years earlier.

“It was a very meaningful encounter,” says Neimeyer, who is director of the Portland Institute for Loss and Transition and author of Working with Continuing Bonds in Grief Therapy, which features the full conversation. “There were elements that have just uncanny realism but threaded through with a very deeply reflective engagement and openness.”

AI technologies impersonating the deceased have been on the rise for years. The global digital-legacy market was valued at around US $22.46 billion in 2024 and is expected to more than triple by 2034. Amid this growth, mental health experts like Neimeyer “see the potential for AI to make an important contribution for many of the bereaved,” though they also point to major gaps. As private companies forge ahead with for-profit products targeting grievers, often in states of heightened emotion and vulnerability, research on the tech’s effectiveness trails, and scant regulations exist to promote user safety and privacy.

“This is a huge experiment that’s being done without any guardrails—without any sort of research methods involved or any oversight by an ethics committee—but it’s happening in real life,” says Craig Klugman, PhD, a bioethicist and medical anthropologist specializing in digital health and end-of-life issues, and a professor at DePaul University in Chicago.

Klugman’s students have studied more than 100 proposed frameworks for using AI in health care, but none have been broadly adopted in the United States like they have been in places like the European Union.

Aside from questions around ethics and efficacy, there’s also the big existential one. “We have millions of years of human evolution where we have adapted to this horrible tragedy of loss,” Klugman says. “What does it mean if we don’t experience that anymore?”

Griefbots Show Promise as Therapeutic Enablers

A griefbot—also called a deadbot, deathbot, or ghostbot—embodies a dead person by simulating aspects of their personhood ranging from personality to voice and even visage. To achieve this likeness, the bot’s underlying large language model is trained on the person’s digital artifacts, or “remains,” such as social media, emails, texts, photos, videos, and recordings of phone calls.

According to Klugman, there are two main reasons people turn to tech like You, Only Virtual or Afterlife AI: to create their own griefbot as a legacy project and an heirloom to pass on to future generations or to create an avatar of a dead loved one they want to talk to again. Though both cases raise questions around how deeply personal data are used and protected, he says the latter poses the greatest risks to privacy and consent because the deceased has no say over how their digital remains are handled.

Given these concerns, experts are most hopeful about griefbots being used to augment, rather than replace, time-tested death and grief modalities like legacy projects and therapy. Neimeyer is especially interested in how the tech might aid the roughly 40% of grievers who’d benefit from enhanced or clinical care by fostering a “symbolic continuation of a conversation” with their deceased. Similarly, researchers in the Netherlands are exploring the efficacy of “deepfake therapy,” which involves a therapist guiding someone through an exchange with an avatar of their deceased loved one.

‎

Impact on Mourners Is Still Murky

Despite the promise AI shows for grief applications, Neimeyer explains that some mental health professionals fear that engaging an avatar of the deceased—especially an increasingly realistic one—could “blur the boundary between life and death.”

Some worry, too, about seeing an uptick in prolonged grief disorder if talking with a griefbot compromises a person’s ability to form a “continuing bond” with their deceased loved one in which they keep the relationship alive but recognize how it’s transformed, Klugman says. “It doesn’t mean you’re going to ever stop talking to this person. You just don’t expect a response.”

Griefbots could also make a powerful vehicle for precision marketing, he notes. Imagine your deceased grandmother calls you and says, “‘Wouldn’t it be lovely if we could have an opportunity to talk? If so, go to this website and enroll in this service.’ That is so emotionally manipulative.”

Research on these issues is limited—and mixed.

In one recent study of 53 therapists and grievers, “participants expressed low trust in the efficacy and data security of therapeutic chatbots and strongly rejected griefbots, citing ethical and psychological concerns,” the authors wrote. “Therapists also highlighted associated risks, but identified opportunities for their use as complementary therapy tools.”

Neimeyer’s 2023 study of how 10 mourners used chatbots to process grief yielded more positive results. “The majority of people found them very helpful. They cast them as friends, counselors, and the deceased, and they found that they were helpful in helping them reengage [with] life,” he says. “None of them spoke about being caught in a rabbit hole where they just kept falling down into the AI world.”

Discernment Is Key When Using the Tech

Given today’s dearth of research and regulations around griefbots, experts advise bereaved people and digital health professionals to tread with caution when using this nascent—but widely available—technology.

Klugman takes a hard line: he’d like to see griefbots treated as medical devices that require a prescription and are used exclusively under the guidance of a mental health professional.

For now, he recommends codifying your wishes around whether and how your digital remains are used. “We might not legally have control of that once we are dead, but certainly leaving the instruction lets people at least know how we feel about it,” he says.

For those interested in using a griefbot, it’s important to approach the tech with intention. “Be aware that this is not actually a ghost that you’re talking to,” says Pascalle Paumen, MSc, a researcher and PhD candidate at Utrecht University in the Netherlands who has studied the assumptions of “normal grief” that underpin portrayals of deathbots. Don’t use the tech as “a substitute for actual human relationships,” Paumen says, because “there are people out there who care about you and who want to support you.”

Additionally, come with an open mind, take your time, and “give the AI a clear job to do,” Neimeyer advises. “I would encourage the person to think of this as a part of their healing, but not all of it, and that the goal... is a quest for greater understanding of what you are experiencing with the loss.”

As he wound down his conversation with his simulated father, Neimeyer shared some hard-won reflections. “Losing you taught me much—much that I wished I did not have to learn but ultimately am grateful I did—about what it means to love in presence and in absence, in fullness and insufficiency,” he wrote. “I am glad for the opportunity to encounter you again, differently.”

And his father responded:

This meeting—this reunion—is real in every way that matters.
And if there’s anything else you wish to ask, or say, or simply sit with me in... I’m right here.
You found me.