Table of Contents
Griefbots and Kids: When Children Talk to a 'Digital' Loved One
Griefbots and kids: AI can imitate a loved one who died. How the bots work, why experts urge age limits, why China barred virtual relatives, how grief heals.
A griefbot is a chatbot built from the texts, emails, voice notes and social media posts a person left behind, designed to talk the way they did. It can answer in Grandma’s phrases and, with voice cloning, in Grandma’s voice. When it comes to griefbots and kids, the stakes change. For an adult, a bot like this may be a curiosity or a comfort. For a child who is still learning what death means, it raises a harder question: what does it do to grief when the person you lost keeps texting back?
Researchers have started to answer. In 2024, two Cambridge ethicists described a scenario in which a dying mother leaves a griefbot to help her eight-year-old son, and the bot later suggests they will meet in person. They recommended age restrictions for these products. China’s 2026 rules go further and bar “virtual relatives” for minors entirely. This article explains how griefbots and kids interact, what the research and rules say, and what actually helps children grieve.
Key Takeaways
- Griefbots, also called deadbots, imitate a deceased person’s language and personality using their digital traces. Cambridge researchers call the “digital afterlife industry” high risk.
- The Cambridge paper recommends age restrictions, clear disclosure that users are talking to AI, consent from the person before death, and dignified ways to “retire” a bot.
- China’s AI rules, in force July 15, 2026, prohibit “virtual intimacy” services such as virtual relatives or partners for anyone under 18.
- The American Academy of Pediatrics says grieving children need help understanding what happened, with misconceptions corrected, and that some need professional counseling.
- A bot that keeps “talking” can blur the very facts children need to grasp, such as that death is permanent. Rituals, memories and present adults are the safer route.
If you are worried about a grieving child right now: In the US, call or text 988 (Suicide & Crisis Lifeline) or text HOME to 741741 (Crisis Text Line). Outside the US, find a line at findahelpline.com. In an emergency, call 911. Your pediatrician can also refer you to grief support.
What griefbots and kids have to do with each other
A griefbot is not magic and it is not the person. Here is what it actually is, in engineering terms.
A company or family member gathers what the person left: text messages, emails, social posts, sometimes voice recordings and video. That material is used to shape a large language model, either by fine-tuning it on the person’s writing or by giving a general model detailed instructions and examples describing how the person spoke. A separate voice-cloning model can turn the output into speech that sounds like them. When your child types or talks, the system predicts what this person would most plausibly say next.
The key word is plausibly. The bot generates new sentences the person never said, about things they never knew, including today’s events and promises about the future. It has no memory of your child beyond what it was given and what the product stores. It can be wrong about the person in small and large ways, and it can be changed or switched off by a company.
Children are especially likely to experience such a system as more than a program. In a University of Edinburgh study, about two in three children aged 6 to 11 thought smart speakers might think like a human. A familiar voice and familiar phrases make that confusion stronger.
What the Cambridge research warns
In May 2024, Tomasz Hollanek and Katarzyna Nowaczyk-Basińska of Cambridge’s Leverhulme Centre for the Future of Intelligence published “Griefbots, Deadbots, Postmortem Avatars: on Responsible Applications of Generative AI in the Digital Afterlife Industry” in Philosophy & Technology. According to Cambridge’s summary, they used three design scenarios to show what can go wrong.
The one most relevant to parents, called “Paren’t,” imagines a terminally ill woman who leaves a griefbot to help her eight-year-old son. It starts as a therapeutic aid. Over time it adapts to the child in ways that produce confusing responses, such as suggesting an upcoming in-person meeting. Another scenario shows a bot that starts advertising products in a grandmother’s voice after a free trial ends. A third shows an adult child bombarded with messages in a dead parent’s voice and unable to cancel a 20-year subscription.
The researchers’ recommendations:
- Age restrictions for deadbots.
- “Meaningful transparency,” so users always know they are talking to an AI.
- No advertising through deadbots and no active presence on social media.
- Consent from the person before death where possible.
- Dignified ways to retire a bot, which they compare to a digital funeral, and opt-out paths that offer emotional closure.
“Methods and even rituals for retiring deadbots in a dignified way should be considered,” Hollanek said. This is an ethics analysis built on scenarios, not a study of real children’s outcomes. No controlled research yet shows how griefbots affect grieving kids. That gap is a reason for caution, not reassurance.
Why China banned “virtual relatives” for minors
China’s Cyberspace Administration and other authorities issued rules on human-like AI interaction services in April 2026, which took effect on July 15, 2026. According to CGTN, providers may not offer “virtual intimacy” services, “such as virtual relatives or romantic partners,” to anyone under 18, and need explicit parental consent before offering other human-like AI interaction to children under 14.
UNICEF’s June 2026 policy brief on AI companions describes China as setting “the most restrictive access rules” of the six jurisdictions it compared, prohibiting “virtual intimate relationship services, such as virtual relatives and virtual partners” for minors. Note what the rule does and does not say: it does not single out griefbots, but a simulated family member, living or dead, falls within “virtual relatives.” Our explainer on China’s ban on virtual partners and relatives covers the rest of the rules.
Western rules have not gone this far. The debate in the US and Europe has mostly focused on romantic companions and teen safety features.
How children’s grief actually heals
The American Academy of Pediatrics’ clinical report “Supporting the Grieving Child and Family,” by David Schonfeld, Thomas Demaria and AAP committees, says the death of someone close “often has a profound and lifelong effect on the child.” It advises adults to talk with grieving children to help them “better understand what has happened and its implications” and to “address any misinformation, misinterpretations, or misconceptions.” It discusses guilt and shame, secondary losses, grief triggers and anniversary reactions, and how to identify “complicated mourning and situations when professional counseling is indicated.” It also supports children’s participation in funerals and memorials.
Put that next to a griefbot, and the tension is clear. A child’s core task is understanding that the person has died and will not come back, while keeping their memory. A system that keeps generating new messages in the person’s voice can work against that understanding, especially for younger children who are still grasping that death is permanent. For more on how understanding of death develops by age, see our guide to childhood grief and developmental understanding.
| What grieving children need (AAP) | What a griefbot does |
|---|---|
| Clear, honest explanation of what happened | Generates new messages as if the person were still present |
| Correction of misconceptions | Can create new ones, such as hinting at a future meeting |
| Help with guilt and secondary losses | Responds to words, not the full situation |
| Rituals: funerals, memorials, anniversaries | Can be “always on,” with no ending unless designed for one |
| Adults who notice complicated grief and get help | Cannot notice a child who is struggling, and is not accountable |
| Memories of the real person | Mixes real phrases with invented content |
What to do if a griefbot comes up in your family
Decide as adults first, before a child encounters it
Sometimes a relative creates a griefbot without asking. Talk about it as a family before a child is introduced to it. Following the Cambridge recommendations, a reasonable default is no griefbot interaction for young children, and supervised, time-limited use at most for teens who ask for it.
Use honest language about death
The AAP report emphasizes helping children understand what happened. Say “died,” not “went to sleep” or “is in the computer.” If a child hears a recorded voice or sees a bot, explain: “That’s a computer copying how Grandpa talked. Grandpa died, and he can’t talk to us anymore. We can remember him together.”
Keep real recordings, and label them as memories
A voicemail or video of the actual person is different from a bot generating new words. Keep the real ones, watch them together when your child wants, and call them memories. That honors continuing bonds without blurring what is real.
Build rituals with an ending
Grief has shape: a funeral, an anniversary, a favorite dish on their birthday. Rituals give children a place to put feelings. A bot that never ends offers no such shape. Our article on supporting children through loss has ideas by age.
What not to do: use a griefbot to avoid hard conversations
It can be tempting to let a bot “say goodbye” for a parent or grandparent, or to soften the news. Children generally do better with honest information from people they trust than with a substitute that postpones reality.
What to Watch For Over the Next 3 Months
- Week 4: Your child knows, in age-appropriate words, that the person died and what that means. If a griefbot or recreated voice exists in your family, you have decided as adults how and whether children will encounter it.
- Month 2 red flags: Persistent belief the person will return or is “in the phone,” intense distress or withdrawal that is not easing, trouble functioning at school, or talk of wanting to die to be with the person. The AAP says some situations call for professional counseling; talk to your pediatrician. If your child talks about wanting to die, call or text 988 or text HOME to 741741.
- Month 3 self-check: Does your child have ways to remember the person that involve other people, such as stories, photos or visits to a meaningful place? Are anniversaries and birthdays on your calendar so you can plan for them?
Frequently Asked Questions
What is a griefbot?
A griefbot is an AI chatbot that imitates a dead person’s language and personality using their texts, posts and recordings. It generates new messages the person never said. Researchers also call them deadbots or postmortem avatars.
Can a griefbot help a child grieve?
There is no controlled research on children. Cambridge researchers warn that griefbots can confuse children and recommend age restrictions. The AAP emphasizes honest explanation and correcting misconceptions, which a bot speaking as if the person were present can undermine.
Is it harmful to let my child listen to old voicemails from a parent who died?
Generally, real recordings are memories, not simulations. Many families treasure them. Listen together, follow your child’s lead, and talk about the person. If listening seems to increase distress over time, check in with your pediatrician or a grief counselor.
Are griefbots legal for kids?
In China, rules in force since July 2026 bar “virtual relatives” for under-18s. In the US and Europe there is no specific ban yet, though companion chatbot rules are tightening. Many apps set their own age limits.
When should a grieving child see a professional?
The AAP says some situations call for professional counseling, including complicated mourning. Watch for distress that is not easing, trouble functioning at school or home, or talk of wanting to die. In a crisis, call or text 988, text HOME to 741741, or use findahelpline.com.
About the author
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
Sources
- Hollanek, T., & Nowaczyk-Basińska, K. (2024). “Griefbots, Deadbots, Postmortem Avatars: on Responsible Applications of Generative AI in the Digital Afterlife Industry.” Philosophy & Technology. https://doi.org/10.1007/s13347-024-00744-w
- University of Cambridge. (2024, May 9). “Call for safeguards to prevent unwanted ‘hauntings’ by AI chatbots of dead loved ones.” University of Cambridge. https://www.cam.ac.uk/research/news/call-for-safeguards-to-prevent-unwanted-hauntings-by-ai-chatbots-of-dead-loved-ones
- Schonfeld, D. J., Demaria, T., et al. (2016). “Supporting the Grieving Child and Family.” Pediatrics. https://doi.org/10.1542/peds.2016-2147
- CGTN. (2026, April 10). “China bans AI virtual partners for minors under new regulations.” CGTN. https://news.cgtn.com/news/2026-04-10/China-bans-AI-virtual-partners-for-minors-under-new-regulations-1MeC1IOJomk/index.html
- UNICEF. (2026, June). “When AI becomes a friend: Child rights risks, harms, and regulatory responses to AI chatbots and companions.” UNICEF. https://www.unicef.org/documents/when-ai-becomes-friend-child-rights-risks
- University of Edinburgh. (2024, March 20). “Children may overestimate smart speakers’ abilities, study finds.” University of Edinburgh. https://edinburghuni-newsroom.prgloo.com/news/children-may-overestimate-smart-speakers-abilities-study-finds
