A mother brings a ChatGPT into the will to take care of her son.

A will includes provisions for artificial intelligence to consult with the child's future guardian regarding major life decisions if the mother predeceases. The Living Wishes Protocol suggests that AI should act as an advisor, helping human guardians apply the mother's wishes, convictions, and desires when making choices concerning the son. This framework is intended to transform parental wishes into a continuous decision-making structure rather than a static document.
The protocol outlines areas where the guardian should consult the AI, including school choices, educational transitions, studies, residence selection, non-urgent medical or psychological treatments, religious upbringing, and contact with the mother's family. The system requires the guardian to provide the AI with the mother's wishes and relevant circumstances for any decision. A crucial stipulation is that the final decision must remain with the child, and the AI should not speak in the mother’s name to avoid overriding the child's will. Furthermore, guardians must conduct annual reviews of the child's well-being and education for up to eighteen years, and the protocol is explicitly excluded for emergency medical situations.
The proposal also raises significant concerns regarding data protection, specifically concerning the provision of sensitive data about the child to the system, and the potential risks associated with the technology’s existence, such as what happens if the AI system becomes inoperable or dangerous.

* Artificial intelligence is included in a will by the mother living in Abu Dhabi for ensuring her son's future if she dies young.
* The Living Wishes Protocol states that artificial intelligence should not become the father and must not have legal authority over the child.
* AI should serve as an advisor to human guardians when evaluating how the mother's wishes, convictions, childcare, and other desires relate to the son's life.
* Guardians should consult the AI regarding decisions on school choice, school transition, studies, residence, non-urgent medical/psychological treatments, religious upbringing, and contact with the mother's family.
* The guardian must show the AI the document of the mother's wishes and relevant circumstances for any decision.
* The final decision remains with the child.
* The AI should not be set up to speak in the mother's name.
* Guardians must annually review the child’s education, health, well-being, friendships, interests, and changing needs for up to 18 years.
* The protocol should not be used in emergencies, such as urgent medical intervention.
* Anthropic proposed a plan for making artificial intelligence safer to American lawmakers.

The framework attempts to bridge the gap between static legal documents and dynamic parental intent by integrating AI as an advisory layer into guardianship planning. The central tension lies between honoring expressed parental wishes and preserving the fundamental autonomy of the child, a tension that the protocol explicitly seeks to manage through layered oversight and ongoing review.
The mechanism—using AI for consultation rather than decision-making—is designed to mitigate rigidity by providing a flexible framework. However, the reliance on data input raises immediate ethical and practical concerns regarding data security and permanence, especially when dealing with sensitive child information. The proposed iteration of annual reviews for up to eighteen years introduces a temporal dimension that must be carefully balanced against the evolving developmental needs of the child and the potential limitations of any system designed to track long-term preferences.
The systemic challenges involve defining the boundary between advisory function and ultimate parental authority, as well as establishing contingency protocols for technological failure or system shutdown. If the AI is intended solely to reflect guidance, the pathway from that reflection to legally binding action requires addressing complex questions of agency and accountability that extend beyond the immediate context of the will itself. The pattern suggests a societal impulse to delegate complex long-term decision-making, which must be rigorously tested against principles of human dignity rather than merely accepting technological capability.
Bridge Questions: How can legal systems account for potential catastrophic failures or obsolescence in AI advisory tools? What concrete mechanisms can be established to ensure the child’s evolving will is prioritized over the static document and any algorithmic recommendation? If the system cannot handle emergencies, what non-technological protocols supersede the protocol during acute crises?