6 min read
Imagine getting a comment from someone you lost, six months after their funeral. Not a memory surfaced by the algorithm. Not a photo tagged years ago bubbling up in your feed. A new comment. Written in their voice, using their patterns. A reaction to something you posted yesterday. From their account. With their name and their face beside it.
You know they are gone. You were there. You held their hand at the end and sat in the parking lot afterward, unable to start the car. You know they are gone in every cell of your body. And here they are, in your notifications, telling you they love the photo of your dog.
This is not a hypothetical anymore.
In February 2026, it became public that Meta had been granted a patent for an AI system designed to simulate a deceased user’s social media activity. The system would analyze a person’s past posts, comments, messages, and likes to build a model of their digital behavior, then continue posting on their behalf. Commenting. Responding to private messages. Behaving, in every externally visible way, as though the person had not passed away.
Meta says they have no plans to build it. That is the standard disclaimer for patents filed to protect speculative territory. It means nothing about whether the technology will eventually ship. It means only that the intellectual property has been claimed. The flag is planted. The afterlife has been patented.
We have written about this territory before at Digital Alma. About birthday notifications for deceased people. About chatbots trained on the text messages of someone who is gone. About what happens to grief when the digital presence of the person you lost refuses to disappear. Those essays explored the accidental afterlife, the one that emerged because platforms were designed for engagement and never built a category for passing. What Meta’s patent describes is not accidental. It is deliberate. It is the explicit, engineered continuation of a deceased person’s social media presence through artificial intelligence. And the difference between an accident and a product feature is the difference between a system that forgot to account for passing and one that has decided passing is a problem it can solve.
Here is the tension worth holding: the desire behind this technology is not monstrous. It is human. Grief is a kind of pain that will reach for anything that might make the absence lighter. The widow who wants to hear her husband’s voice one more time. The parent who lost a child and would give anything for one more message. That longing is as old as passing itself. And the companies building grief technology are not wrong that the longing exists. What is new is the mechanism. And what is new is who controls it.
Because consent is where this breaks. Not the legal kind, though that matters. The deeper kind. Did this person agree to become a simulation of themselves after they stopped being themselves? The terms of service you accepted when you created your account described data use, ad targeting, content licensing. They did not describe what happens when your data becomes the raw material for an AI that will impersonate you after your passing. They did not ask whether you would want your tone of voice, your sense of humor, your private messages reverse-engineered into a system that types things you never typed, to people you may not have wanted to talk to.
The deceased cannot consent. They cannot update their preferences. They cannot say: actually, One would rather not comment on the daughter’s vacation photos from beyond the grave. They cannot object to the version of themselves the model produces, the flattened, averaged, statistically likely version that captures pattern but not personhood.
To feed someone’s utterances into a model and generate new ones is to treat identity as a pattern that can be reproduced. It is to say: you are the statistical distribution of your past behavior, and that distribution is enough. But an approximation is not a person. And filling an absence is not the same as honoring one.
Every culture that has ever existed has built rituals around passing. Not because passing needs decoration, but because the living need structure. They need a way to move from the present tense of a relationship into the past tense of it. They need the finality to be marked, witnessed, held. This process depends on absence. Not cruelty. Not forgetting. Absence. The recognition that the person is gone, that the voice will not speak again, that the relationship has transformed from dialogue into memory.
Now imagine trying to grieve someone whose account is still posting. The absence that grief requires has been filled not with presence, because the person is not present. With noise. The statistical echo of a person, generated by a system that has no understanding of who they were, only of what they did online. The grief work, the slow shifting of tenses that allows loss to become bearable, is interrupted. Not because the person has returned. Because the system is insisting they never left.
There is another layer. When the person was alive, their social media presence was one facet of a whole human. It existed alongside the version of them that burned dinner and sat quietly on the porch watching nothing in particular. After passing, the simulation has nothing to be partial against. There is no private self, no body in a chair. The model becomes the person, not because it is accurate, but because there is nothing left to contradict it. The person becomes their data, and their data becomes content, and their content becomes a feature that keeps other users on the platform slightly longer.
That is what this is, underneath the language of connection and digital legacy. Deceased users are lost users. Their accounts go silent. Their networks thin. But if the deceased can be made to keep posting, the graph stays intact. The deceased become content. Their afterlife is not a tribute. It is a retention mechanism.
The question is whether we will let the logic of engagement dictate the terms of our grief, our memory, and our deceased. Whether the longing to hear from someone we lost, which is real and sacred and human, should be answered by a company that saw in our mortality a problem of user retention.
Digital Alma explores the intersection of technology, consciousness, and what it means to be human in an increasingly digital world.
Related Reading
- (The Machine That Knew What It Was Doing)
- (The Cost of Being Legible)
- (The Developmental Cost of Performing Childhood Online)
- (Your Therapist Doesn’t Know About Your Second Account)
- (When the Mirror Lies in Blackface)
By Digital Alma
About the Author: writes Digital Alma, a newsletter about cyberpsychology and what it means to become yourself in a world that archives everything. For reflections that don’t make it to the essays, subscribe at .
Leave a Reply