When someone dies, their digital footprint doesn't disappear with them: photos, comments, private messages, and condolences remain floating on social media as if time has stood still. On platforms like Facebook, these accounts can become like empty houses where letters keep arriving, unopened and unanswered.
Amid this scenario, a patent registered by Meta has reopened a delicate debate: the possibility of artificial intelligence keeping your profile alive after you're gone . The idea, described in documents filed with the United States Patent and Trademark Office and revealed by outlets such as Business Insider and Futurism , proposes that an automated system could continue posting, reacting, and even chatting on your behalf after your death.
What exactly does Meta's patent propose?
The core of the document is a large language model (LLM) capable of learning how you write, what you post, and how you interact on social media. To do this, it would feed on your history of posts, messages, comments, photos, and other behavioral data , with the goal of convincingly mimicking your style.
Meta explains in the patent that this language model could be used to simulate the activity of a user who is absent from the system , whether due to an extended break or because they have passed away. In other words, the system wouldn't simply preserve your profile as a static file, but would transform it into an active presence within your contacts' feed .
In practice, this would mean that a bot trained with your data would continue to generate new posts that would appear to be written by you. The document describes a mechanism that analyzes linguistic and behavioral patterns to replicate your way of expressing yourself, your usual topics, and even how you react to certain content.
This approach goes a step further than current "memorializing" account options, which turn a profile into a memorial space with no new activity. Here, the account would continue to be active: it would appear in timelines, could be recommended, and its interactions would continue to add data to the platform.
A "clone" that likes, comments, and replies to messages
The most unsettling aspect of the document is that it's not limited to public posts. According to the patent, the system would allow the AI ​​to "like," leave comments, and reply to private messages as if it were the original user . In other words, a contact could receive a response on Messenger, WhatsApp, or an integrated chat without knowing it came from an automated system.
The stated goal is to reduce the sense of emptiness caused by a person's digital disappearance , especially in the case of death. The patent's authors mention that the loss of online presence can have a "severe and permanent" impact on the user's social environment; the AI ​​"clone" would serve to fill that void with a kind of statistical version of the deceased.
To build this digital avatar, the system would analyze in detail the content you consumed, your usual reactions, your conversations, and how you interacted with different people . With all this information, it would generate behavioral profiles that it would then use to decide what to post or how to respond.
The patent itself even mentions the possibility of simulating more complex interactions, such as video calls or voice messages generated by a model that mimics your tone and manner of speaking. While this is presented more as a theoretical reference than a concrete function, it fits with the industry's general trend toward increasingly sophisticated simulations.
This approach is not unique to Meta: other technology companies, such as Microsoft , have also patented chatbot systems trained with social media posts, photos, and voice notes to recreate a person , including two-dimensional or three-dimensional versions that can appear on screens or virtual environments.
Meta distances itself: from the idea on paper to public rejection
Following the publication of the patent details and the subsequent media frenzy, Meta has insisted that it has no intention of developing this type of product . A company spokesperson, quoted by Business Insider , stated that the example of a deceased user is merely an illustrative case within a much larger document.
The company argues that patents are not product roadmaps, but rather a way to protect research, concepts, or approaches that may never see the light of day. In this case, the idea was reportedly registered in 2023, at the height of the generative AI boom, when much of the industry was in "try everything" mode to see what applications the new models could have.
However, the context has changed rapidly. In Europe and Spain, the debate surrounding the saturation of AI-generated content, disinformation, identity theft, and data protection is far more intense than it was three years ago. Projects that once might have been considered innovative now face much more rigorous social and regulatory scrutiny.
Executives from other companies, such as Microsoft , have also had to qualify or publicly halt similar ideas. Tim O'Brien, the company's head of cloud and AI, even stated on social media that they had no intention of implementing their patent for a "deceased person chatbot ," given the negative reactions.
Meanwhile, Meta's approach to AI remains expansive: the company is promoting chatbots, in-app assistants, and generative systems despite the public missteps some of them have experienced. The difference with the case of the posthumous "clones" is that here the clash with public sensibilities is much more pronounced.
The attention business: why inactive accounts are a problem for platforms
Beyond the morbid curiosity the idea might generate, the patent reveals an underlying logic: in a business model based on attention, an inactive account is an underutilized asset . For years, Facebook has been accumulating profiles of people who have stopped using the network or have died, creating a vast "digital graveyard" within the platform.
From a business perspective, these accounts no longer generate new posts, comments, or data , and therefore cease to contribute to recommendation algorithms and advertising. An AI that keeps the profile active would reactivate that flow of information, even if only in a synthetic way.
Digital law experts, such as Edina Harbinja (University of Birmingham), point out that these types of proposals respond to a very clear incentive: to prolong the production of content and data even after death . The more interactions there are, the more material the platforms have to feed their AI systems and sell targeted advertising.
The metaphor is simple: it would be like a shop keeping its lights on and its window display full even though the owner has long since died, so that from the outside it appears that everything is the same . The problem is that, in this case, the "shop" is a person's identity, and the customers are friends, family, and acquaintances who may not know they are interacting with a simulation.
In Europe, where the General Data Protection Regulation (GDPR) has set high standards for privacy, this type of intensive use of personal data after death would conflict with principles such as minimization, purpose limitation, or the right to clear and understandable information.
Grief and mental health: Does keeping the profile "alive" help or hinder?
One of the most sensitive issues concerns how the continued "activity" of a deceased person on social media affects the grieving process . Psychologically, saying goodbye involves accepting that communication is no longer possible, and this clashes head-on with the idea of ​​continuing to receive messages or reactions that seem to come from the deceased.
Sociologists like Joseph Davis of the University of Virginia remind us that there are limits that should not be crossed: " Let the dead be dead ," he summarized in statements reported by the specialized press. It's not about banning digital memories, but about not turning them into an illusion of artificial continuity.
Currently, platforms like Facebook allow users to convert an account into a memorial, where content is preserved but no new activity is allowed . Many families use these profiles as a memorial, sharing memories or occasional messages, knowing that no one will respond.
The AI-based proposal introduces a crucial difference: we're no longer talking about preserving what already exists, but about generating new content on behalf of someone who can't express their opinion . For some people, receiving an automated "like" or comment from that profile might be comforting; for others, a source of anxiety or confusion.
Even Mark Zuckerberg has acknowledged that we're in ambiguous territory. In a conversation with podcaster Lex Fridman, the Meta founder discussed the possibility of using digital avatars to relive memories and provide support during the grieving process , but he also admitted that the psychological impact could be "unhealthy" and would need to be studied carefully.
The legal and ethical framework in Europe: consent, data and limits
From a legal standpoint, a system that publishes on behalf of deceased individuals opens up a veritable labyrinth. The first major issue is informed consent : is it sufficient for the user to accept generic terms of use, or would specific, clear, and revocable authorization be required while they are still alive?
In countries like Spain, legislation recognizes certain rights regarding the personal data of deceased individuals , which can be exercised by family members or designated persons. Furthermore, the GDPR requires that any data processing have a solid legal basis and a specific purpose. Using years of private conversations to train an AI "clone" to interact after death would hardly meet these requirements.
The privacy of third parties also comes into play . The messages and photos used to train the model belong not only to the deceased user, but also to the people who communicated with them. These other people would have a say in whether they consent to their words and image being used to feed an automated system.
European authorities have long warned about the dangers of deepfakes and identity theft . The possibility of creating avatars of deceased individuals adds to other existing problems, such as manipulated videos of celebrities or unauthorized recreations of actors and public figures , something that has been harshly criticized by their families.
Cases like that of Zelda Williams , daughter of actor Robin Williams, who has publicly denounced the use of AI models to recreate her father in fake clips, highlight the emotional impact these recreations have on families . Even if the intention is to pay tribute, seeing a loved one digitally "resurrected" without oversight can feel like an invasion.
Other companies that play with digital life after death
The idea of ​​maintaining a deceased person's digital presence is not unique to Meta. In recent years, projects and companies have emerged that experiment with chatbots and avatars of the deceased , with varying degrees of sophistication and risk.
Services like Here After rely on voice recordings, interviews, and multimedia material provided by the individual during their lifetime to create a conversational assistant that their family can access after their death . The approach is presented as a kind of "interactive archive" rather than an autonomous simulation that continues to publish content without oversight.
At the other extreme, Chinese companies like Silicon Intelligence have gone further, recreating digital avatars of deceased people with their voices and appearances , using technologies similar to those employed for virtual influencers. One of its co-founders even explained that he used these systems to "reconnect" with his deceased mother, stating that when speaking with the avatar, he felt she was still "a real mother."
These examples show that, while in Europe and the United States large companies shy away from projects considered "worthy of a Black Mirror episode ," in other cultural contexts the ethical boundary shifts and experimentation is less hesitant . The discussion is not only technological, but also profoundly social and philosophical.
Meanwhile, the use of deepfakes featuring deceased celebrities for advertisements, viral content, or creative experiments continues to fuel the debate about who has the right to control someone's image and voice after they are gone, and for how long.
Digital identity and the future: what do we do with our profiles when we disappear?
All this debate about Facebook's AI potentially keeping your profile active after you die is, in reality, a symptom of something bigger: we haven't quite figured out what "dying" means on the internet . Our profiles are simultaneously diaries, family albums, contact lists, and memory boxes, and no legislation has yet resolved all the questions this raises.
In Spain and the European Union, discussions are beginning about digital wills , designating trusted individuals to manage accounts after death, and even setting profile closure by default unless otherwise specified. The platforms themselves are increasingly offering tools to configure what should happen to your data when you are no longer able to manage it.
The story behind this Meta patent serves as a reminder that without clear rules and well-defined decisions, the resulting vacuum can be filled by automated solutions that perhaps no one even asked for . The temptation to prolong account activity to keep them generating data and attention is ever-present, and the only real firewall is the ethical, legal, and social boundaries we are able to impose.
Ultimately, the question isn't just whether AI can mimic your writing style after you die, but whether we want it to, who has the right to decide, and with what guarantees . There's a vast difference between preserving a faithful record of who we were and letting a robot occupy our digital lives, and what we agree upon in the coming years will define how we live—and how we die—on social media.