Meta Tests AI Agents for Deceased Users
Meta is reportedly experimenting with AI agents that could manage a person's Facebook or Instagram account after they die. The agents could maintain a digital presence or carry out legacy wishes, raising new product and ethical questions about how users will want their digital footprints handled posthumously. This signals that memory and legacy are becoming active product features in social tech.
This concept isn't entirely new for Meta; a patent granted to the company outlines an AI that could simulate a user's social media activity after they die. The patent, first filed in 2023 by CTO Andrew Bosworth, describes training a language model on a user's historical posts, comments, likes, and even private messages to create a "digital clone." Internally, a similar idea was reportedly dubbed "Project Lazarus" three years ago. The recent patent even suggests the AI could simulate audio or video calls, aiming to lessen the "severe and permanent" emotional impact on followers when a user passes away. Despite the patent, Meta has stated it has "no plans to move forward with this example." This potential feature goes far beyond Facebook's current "memorialization" option, first introduced in 2009. Today, a memorialized account displays "Remembering" on the profile and prevents logins, while a designated "Legacy Contact" can manage tribute posts but cannot read messages or post as the deceased. The experiment places Meta within a growing "grief tech" or "death tech" startup scene. Companies like Eternos, StoryFile, and HereAfter AI are already building products that create interactive AI avatars from a deceased person's digital footprint. Many of these companies were founded by engineers and entrepreneurs after experiencing personal loss. A key technical and ethical challenge is obtaining explicit, pre-mortem consent from users for how their data will be used. Researchers also warn of the psychological risks to the bereaved, such as hindering the grieving process or creating unhealthy attachments to a "deadbot." For Meta's engineering teams, this signals a broader strategic shift. The company has been downsizing its foundational AI research division (FAIR) to concentrate resources on a new "Superintelligence" product-focused group. This move prioritizes engineering-driven product velocity over more open-ended research, a cultural shift relevant to engineers considering their career paths in big tech versus startups.
