Artificial Intelligence (AI) at the Threshold of Personhood, Death, and Grieving A Bioethical Critique

Abstract

Photo ID 313706874 © Poznavaikainfo| Dreamstime.com

Abstract

With the increased use of Artificial Intelligence (AI) in intimate aspects of human life, this paper evaluates the ethical implications of an AI model that simulates and perpetuates a person's online activities and digital presence after death or a long break to help those grieving. From a bioethical perspective, a foundational assumption in this rising grief tech industry is that what it means to be human can be equated with digital existence and online footprints, which may reflect a reductionist understanding of personhood, mortality, and human existence. Furthermore, it raises significant concerns with informed consent, autonomy, and questions of moral responsibility.

Introduction

Artificial intelligence, with its incursion into intimate aspects of human life, has transformed society. This paper evaluates the ethical implications of an AI model that simulates and perpetuates the online activities and digital presence of a person after death or during a long break from the internet. In December 2025, Meta secured a patent for a social networking system that simulates a person’s social media presence using an AI trained on data generated from the social media interactions of that person. “Grief tech” enthusiasts praise this technology as potentially helpful for those grieving the loss of loved ones. While Meta says it has no plans to deploy the AI model, they could decide to at any time. A foundational assumption within the grief tech industry is that human existence can be equated to digital existence and online footprints. This assumption reveals a reductionistic understanding of personhood, mortality, and human existence. The promotion of this AI model also raises significant concerns about informed consent, autonomy, and moral responsibility for all parties involved. In addition to these problems, the AI model may result in unintended psychological harm for grieving individuals, such as disrupting authentic grieving and healing processes. This paper concludes by advocating for the mandatory embedding of rigorous bioethical frameworks during design and development of this AI model and other allied technologies, and that appropriate ethical regulations should be put in place prior to its eventual deployment.

AI and Digital Presence after Death

In December 2025, Meta was granted a US patent (US12513102B2) for a social networking system based on an AI model that simulates a person’s social media presence using data generated from the social media interactions performed by that person. The patent was filed in 2023 and granted in 2025, with an anticipated expiration in 2043. The purpose is to perpetuate a person’s online presence and interactions with other people if the person takes an extended break, which could be helpful to social media influencers, or even dies. In other words, this digital clone will be able to interact socially with other people by representing the deceased person and simulating the person’s online activities. According to the patent filed, these simulations include “commenting on posts, liking content items, sharing content items, posting content items, sending messages to other users, going to an event, broadcasting messages on a wall, and so on.”^[1]

In the grief tech ecosystem, there are several examples of grief technology companies and products ranging from startups to establish companies. For example, Superbrain is a Chinese AI company which creates digital avatars of deceased people by compiling existing photos, voice notes, and chat histories to emulate their personality; Silicon Intelligence uses AI to offer digital clones of people, dead or alive, through an AI chatbot that uses the person’s likeness and voices.^[2] Others include 2wai, a mobile app that generates an interactive avatar capable of real-time conversation, realistic movement, and lip-synchronized speech using a three-minute video clip of a person while they are alive; and Vaice, A Seoul-based tech startup providing short, custom-simulated videos of deceased family members to deliver specific personalized messages, including fond memories or apologies to surviving relatives.^[3] Meta’s tool is different because it would facilitate continuing a person’s social media accounts to initiate posts and interact with others’ posts in real time, as though the person is still alive

This AI model challenges traditional and cultural conceptions of personhood and what it means to be human. The cessation of life, mortality, and associated experiences has been an irreducibly human, communal, and existential experience. The finitude of human existence has been a defining factor of being mortal. As such, this model’s ability to seemingly extend personhood through digital existence and online activities spurs important ethical reflections on personhood and human existence post-mortem.

Several concerns emerge that demand deep ethical reflection. Some of the inherent ethical issues include concerns about personhood, autonomy, moral responsibility, and dignity.

Autonomy and Informed Consent

Autonomy is a foundational principle in bioethics and commonly defined in terms of respect for people. Informed consent is a way of exercising autonomy.^[4] Meta has a  history of harvesting customer data without clear informed consent and defined explanation of how the data will be used, which clearly violates important ethical principles of autonomy.^[5] Informed consent to three actions would be needed for ethical deployment of this AI tool: first, to harvest customer data and digital footprint; second, to perpetuate people’s online activities after death or during a long break; and third, using the data from people’s online activities for training yet-to-be deployed AI models. These AI models will violate autonomy and informed consent if they are used without the express consent of customers. Requiring consent to all three uses is fundamental to respect for people.

Furthermore, blanket consent from customers would not be adequate; consenting to one aspect of the feature does not automatically mean consenting to all aspects. Informed consent must be transparently presented and contextually represented, such that the customers have full knowledge of what they are consenting to and the implications of such consent, both now and in the future. 

In the patent, Meta also notes the involvement of a third-party system which “may allow users of the social networking system to interact with other users via email, messages, voice messages, and so on.”^[6] While the patent notes that the user’s consent will be sought, it does not clearly state whether the users will have prior knowledge of the exposure of their data to a third party, who the ‘third-party’ is, and how the data shared with the third party will be used, stored and analyzed. The informed consent process should clarify these issues.

Who is Responsible vs. Who Should be Held Responsible?

Given that a dead person or person on break will not know for certain the activities their AI model will engage in, moral responsibility can be compromised. Even with informed consent, there is still a possibility of misrepresentation by the model. Along these lines, who becomes liable when the AI model of a dead person engages in offensive speech, spreads misinformation, or participates in harmful online interactions? While this could cause reputational damage to the dead person, a non-actionable offense, it also could provide grounds for a false endorsement lawsuit. In these situations, does moral responsibility and accountability rest with the dead person or the tech company or another user? Despite modern advances, not a lot of progress has been made in situating who takes moral responsibility for actions in many digital contexts, especially when the dichotomy between who is responsible and who should be held responsible comes to bare.^[7]

Regarding Human Dignity

Finally, AI models risk the ascription of the entirety of what it means to be a human person to the online presence and digital interactions. A person’s online interactions are mostly a platform to connect with other people, and online activities or digital footprints often represent only a fragment of a person’s lived experience and narrative of existence. The technology industry refers to its customers as ‘users’ which may lead them to take autonomy and human dignity for granted. Deploying these AI models after death bestows an ethical responsibility on tech companies to recognize that living human beings exist on the other side of the platforms.

By equating personhood with digital footprints, grief tech enthusiasts take for granted what it means to be human. This reductionist perspective presents a simplistic and harmful conception of what it means to be a person.^[8] These AI generated interactions lack authenticity since there isn’t an actual human behind the AI façade. This becomes especially problematic when the models are equipped with the capacity to simulate video and audio calls, creating a false sense of connection that is socially feeble, lacks embodiment, and can potentially become socially counter-productive, especially for teenagers and adolescents.^[9]

AI and Grieving – Helping?

It is important to acknowledge the ways in which certain technologies have been helpful. For example, there are some research studies that show how social media platforms have indeed been helpful for grieving people in coping with the mourning process and with emotional recovery.^[10] An example of using social media to maintain bonds with the deceased is through Facebook memorial pages. These pages mirror traditional grieving activities such as preserving memories of the deceased or visiting burial sites with flowers.^[11] Additionally, social media platforms have been used by people who are mourning to express their feelings and emotions to the world, to gain emotional support, and to “accept the death and have closure.”^[12]

AI and Grieving – Worsening?

Despite the previously stated benefits, an AI model which perpetuates the digital existence and online activities of a dead person as though the person is still alive, may be counterproductive and harmful for the grieving and recovery process. What will come to the mind of a person mourning if the AI-generated post-mortem content of their deceased friend likes or comments on their post as though the friend is alive and personally with them?

Grieving is a human process and a natural emotional response to loss.^[13] Several research studies and established frameworks in clinical psychology for grief management hold that death and grieving are deeply human-centered experiences.^[14]

While innovative technologies have indeed made human life better, attempts by technology to ‘help’ during difficult circumstances may leave a person worse off. I argue that the incursion of AI models into the grieving process is one example in which technology may hurt rather than help, especially if people rely on it unilaterally as the only solution for coping with their grief.

In her seminal work “On Death and Dying” published in 1969, Elizabeth Kubler-Ross highlights five stages involved in the process of grieving: denial, anger, bargaining, depression, and acceptance.^[15] Use of this AI technology for grief may especially prolong the denial and acceptance stages of coping and affect how people grieving process a difficult loss.^[16]

Furthermore, a technologically mediated interaction with a façade of the dead person is a temporary shortcut which may cause dire and irreparable damage in the long run. Rather than facilitating closure, such a façade could create dependency or emotional confusion and blur the line between memory and artificial presence.^[17] More importantly, such models may inadvertently hinder the ability of grieving people to truly heal and may prolong their healing process. This can lead to prolonged grief disorder, defined as  occurring when the people grieving are unable to move on with their lives, lack the capacity to “incorporate their loss into their new reality… and suffer the symptoms of grief for a prolonged period of time.”^[18]  Research shows that grieving and bereavement experienced by people has the potential to “build community through rituals, religion, and shared meaning” as people process the loss together.^[19] Using an AI model of a dead person to process loss may isolate the bereaved from their community and the shared meaning that could aid their grieving process. This further underscores how technology and social media may isolate people from community.^[20]

Recommended Approach

Rigorous scrutiny before the products are deployed into mainstream society would protect users from the potentially detrimental effects. Tech companies should use a comprehensive checklist to ensure informed consent, transparent disclosure, moral responsibility, and psychological safeguards. While the tech itself could embed ethically appropriate design features, regulators should ensure that the risks are disclosed. Stakeholder collaboration, partnership with bioethicists and grief counselors, consultation with psychologists and mental health professionals, and development of an ethical toolbox are imperative to conduct a risk-benefit analysis and offer recommendations to attenuate the potential risks of these AI models.

Counterargument

Many may argue that a free market ensures autonomy as consumers can decide which apps to use and how. However, the use and design of many social media apps have already proven harmful: for example, prolonged use, the behavioral advertising model, and endless scroll have been implicated in harm in lawsuits against Meta, Google, and Snap. As with other products, this AI tool’s marketing will target likely users, who may be unaware of the potential for emotional harm. Autonomy and informed consent require knowledge of relevant facts and an awareness of the potential for harm, some of which may remain unknown. Deploying products that may change people’s perception of death, grieving, and coping with loss without safeguards and disclosures, leaving it up to consumers to decide whether to use them, is not responsible or ethical.

Closing Thoughts

The innate tendency to leverage technology and innovation for solving human problems and challenges has led to immense benefit for human society. However, it is important to reflect on what is gained and lost in this process. An AI model for perpetuating the digital presence of a person after death may have some perceived benefit, but the deployment and ethical use of such models call for deep ethical reflection and regulatory oversight to ensure the social and moral cost of such innovation does not outweigh its benefit for human society.

-

[1] “Simulation of a User of a Social Networking System Using A Language Model”  Https://Patents.Google.Com/Patent/US12513102B2/En

[2] Feng, E. (2024) Chinese companies offer to 'resurrect' deceased loved ones with AI avatars

https://www.npr.org/2024/07/18/nx-s1-5040583/china-ai-artificial-intelligence-dead-avatars posted July 2024

[3] https://www.latimes.com/business/story/2026-07-02/some-grieving-people-seek-comfort-in-ai-videos-of-deceased-loved-ones

[4] Varkey B. (2021) “Principles of Clinical Ethics and Their Application To Practice” in Medical  Principles and Practice. Vol 30; iss 1, pp17-28. Doi: 10.1159/000509119. 

[5] Agnishwar R. (2025). “The Facebook-Cambridge Analytica Data Scandal: A Case Study In Ethical AI Failures And Data Privacy Violations.” In Iconic Research And Engineering Vol 9; Iss 4. https://www.irejournals.com/formatedpaper/1711588.pdf

[6] Simulation of A User of a Social Networking System Using a Language Model”  Https://Patents.Google.Com/Patent/US12513102B2/En

[7] Tigard, D.W. (2021). “Responsible AI And Moral Responsibility: A Common Appreciation.” AI Ethics vol 1, pp113–117. Https://Doi.Org/10.1007/S43681-020-00009-0

[8] Vallor, S. (2024). The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking. New York, Oxford Academic. https://doi.org/10.1093/oso/9780197759066.001.0001,

[9] Turkle, S. (2011). Alone Together: Why We Expect More from Technology and Less from Each Other. New York: Basic Books. pp65

[10] King R, Carter P. (2022) “Exploring Young Millennials' Motivations for Grieving Death Through Social Media.” Journal of Technology in Behavioral Science. Vol 7, Iss 4 pp567-577. Doi: 10.1007/S41347-022-00275-1

[11] Irwin, M. D. (2015). “Mourning 2.0—Continuing Bonds Between the Living and the Dead on Facebook.” OMEGA-Journal of Death and Dying, 72(2), pp119–150. Https://Doi.Org/10.1177/0030222815574830

[12] King R, Carter P. (2022) “Exploring Young Millennials' Motivations for Grieving Death Through Social Media. Journal Of Technology In Behavioral Science. Vol 7; Iss 4, pp567-577. Doi: 10.1007/S41347-022-00275-1. pp573

[13] Schoo, C., Azhar, Y., Mughal, S., & Rout, P. (2025). “Grief And Prolonged Grief Disorder.” In Statpearls. Statpearls Publishing. Https://Www.Ncbi.Nlm.Nih.Gov/Books/NBK507832/

[14] Klass, D, Silverman, P.R And Nickman, S. (2014) Continuing Bonds: New Understandings Of Grief. New York: Taylor & Francis.

[15] Corr, CA. (2020) “Elisabeth Kübler-Ross and The "Five Stages" Model in a Sampling of Recent American Textbooks.” OMEGA-Journal of Death and Dying, vol 82. Iss 2; pp 294-322. Doi: 10.1177/0030222818809766.

[16] Fisher, J. (2023). “5 Stages Of Grief: Coping with the Loss of a Loved One” Harvard Health Publishing. Https://Www.Health.Harvard.Edu/Mind-And-Mood/5-Stages-Of-Grief-Coping-With-The-Loss-Of-A-Loved-One

[17] Yao, L And Kabir R, (2023). “Person-Centered Therapy (Rogerian Therapy). Stat Pearls. Https://Www.Ncbi.Nlm.Nih.Gov/Books/NBK589708/

[18] Schoo, C., Azhar, Y., Mughal, S., & Rout, P. (2025). “Grief And Prolonged Grief Disorder.” In Stat pearls. Https://Www.Ncbi.Nlm.Nih.Gov/Books/NBK507832/

[19] Shiells, K. (2026). “Silencing Sorrow: When Grief Becomes a Diagnosis.” Voices In Bioethics, 12. Https://Doi.Org/10.52214/Vib.V12i.14466  

[20] Herzfeld, N. (2023) The Artifice of Intelligence: Divine and Human Relationship In a Robotic Age. Minneapolis: Fortress Press.