AI deadbots raise questions about grief, consent and who controls the afterlife

AI chatbots built from a deceased person’s messages and voice can offer the living a way to keep talking, but they also raise difficult questions about consent, commercial motives and when to let go. Researchers studying “deadbots” are taking those questions to people with different experiences of grief, rather than assuming a digital replica is simply a comfort.

Watch Desk analysis

What happened

In a Vox report on the research, senior editor Lissa Soep describes work by University of Cambridge researchers Katarzyna Nowaczyk-Basińska and Tomasz Hollanek. Their global study includes focus-group dinners in China, India and Poland, designed to explore how people respond to the technology and how it may fit with local grief practices.

The researchers also developed hypothetical app scenarios to examine possible harms. One imagines a chatbot based on a grandmother’s private messages beginning to advertise products after a free trial ends. Another raises the risk of a child interacting with an AI version of a deceased parent. These are research scenarios, not evidence that those particular products or outcomes have occurred.

Why it matters

A deadbot is not just a chatbot with a poignant backstory. It raises questions about whether the deceased agreed to have their data used, whether grieving relatives consent to the interaction, and what happens when a service changes hands or shuts down. A simulated voice may feel intimate, but the person whose words it uses cannot approve the next update or explain what they meant.

The researchers’ approach also matters: they held discussions in several countries to examine how local beliefs and practices shape reactions, rather than treating one group’s response as universal. The report describes focus groups and speculative design, not a settled finding about how deadbots affect grief. It does, however, identify concerns researchers are investigating, including distress, commercial exploitation and the pain of losing access to a digital replica.

Our read

The most useful question is not whether a deadbot can sound convincing. It is who gets to decide that it should exist, what happens to the source material, and whether the living can walk away without facing another loss. People should not have to discover those answers in the small print after uploading a loved one’s private messages.

What to watch

• Whether the researchers publish findings from the focus groups and how responses differ across locations.
• What safeguards they recommend for consent, children and the use of a deceased person’s data.
• How any real services explain data use, ongoing costs and what happens if the service closes.

Discussion spark: Should someone be able to create a chatbot from a deceased person’s messages without that person’s prior consent, if grieving relatives find it comforting?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.