27/10/2026
Can an AI help raise a child after a parent's death? Giovanni Spitale on Dubai Eye
Giovanni Spitale, Co-Director of the ITE Lab at the Institute of Biomedical Ethics and History of Medicine, University of Zurich, was interviewed by Georgia Tolley on The Agenda, Dubai Eye 103.8 FM. The topic was a case that has drawn attention in the United Arab Emirates: a will that asks a child's guardian to consult ChatGPT.
According to Spear's, a mother living in the UAE has drafted a will with lawyer Byron James of Expatriate Law. It requires her son's guardian to put every major decision to ChatGPT, including schooling, health, faith and where the child will grow up. The model is guided by a document in which she wrote down her values, principles and wishes. The guardian must log every question, answer and decision. They may depart from the model's answer, but must explain why.
The case closely matches scenarios described in "The Making of Digital Ghosts: Designing Ethical AI Afterlives", which Spitale published with Federico Germani in Ethics and Information Technology in June 2026. The paper proposes a nine-dimension taxonomy of digital afterlife technologies. It names decision support, posthumous proxy agents and systems involving children among the configurations that need the strictest safeguards.
In the interview, Spitale discussed where such an arrangement can help and where it becomes risky. A faithful record of a parent's explicit wishes can be more useful than a static letter. But the decisions the model will be consulted on are the ones the mother never anticipated. On those, it stops reporting her views and starts generating new ones. The system does not pretend to be the mother, so the risk is not impersonation. The risk is authority: if every departure has to be justified, advice becomes something close to a binding instruction, delivered by a commercial product that owes the family nothing.
Reference: Spitale G, Germani F. The Making of Digital Ghosts: Designing Ethical AI Afterlives. Ethics and Information Technology 28:34 (2026). DOI: 10.1007/s10676-026-09910-4 (open access, CC BY 4.0)