AI Griefbots Could Make Grieving Even More Complicated

The Trend: What if you could text someone you’ve lost — and they texted back? That’s the promise of AI “griefbots,” or “thanabots”: systems trained on a deceased person’s texts, emails, photos, videos, social posts, and voice recordings to generate a digital imitation of how they might speak and respond. Generative AI makes these bots increasingly convincing, and startups are already selling versions of the technology.

What People Are Saying: There’s an obvious appeal. Hearing a familiar voice or having a conversation could offer comfort, help someone revisit memories, or provide a way to work through unfinished feelings. Some therapists and researchers see potential for working through grief, but the science is young, and experts are cautious. There are also thorny questions: Did the deceased consent? Who gets to decide what “version” of them the bot creates? And could repeatedly interacting with a convincing simulation make it harder to accept that the person is gone? Researchers have specifically called for consent and safeguards against unhealthy emotional dependence.

What to Know: Grief isn’t a problem to solve, and it doesn’t have a timeline to complete. It’s a process of learning to live with an absence. Continuing bonds with someone who died can be part of that process. AI may eventually have a place in that landscape, but for now, it’s wise to see griefbots as an experimental technology at best.