Artificial intelligence is opening a new frontier in the relationship with death. Photographs, videos, messages, emails, and voice recordings can be used to create digital representations of deceased people, from videos in which they apparently pronounce new messages to avatars capable of answering questions.
These systems are named griefbots, deathbots or posthumous avatars and are part of an industry known as grief tech or digital afterlife. The technological leap is relevant: it is no longer just about animating a photograph, but about recreating voices, memories, and ways of expressing oneself to simulate a conversation with the deceased.
From the United States to China: the business of "digital life"
One of the best-known companies is HereAfter AI, developed in the United States. Its system interviews a person while they are alive and stores stories, memories, and recordings of their voice so that later their relatives can ask them questions and listen to their answers.
StoryFile uses a similar model: it previously records hundreds of video responses and employs artificial intelligence to select the most appropriate one depending on the question. In 2022, its technology allowed British educator Marina Smith, who had recorded her answers before dying, to apparently converse with attendees at her own funeral.
Other companies go further. You, Only Virtual works on digital representations capable of reproducing memories and ways of speaking, while Eternos and Seance AI offer different systems of legacy or digital conversation. Australia has Afterlife AI and in China, Silicon Intelligence has begun to market services presented as "digital resurrection." South Korea also has technologies capable of reproducing the face, voice, and expressions of deceased people.
It is not the same to prepare an avatar as to be recreated after death
There is, however, a fundamental difference. A person can decide in life to record their voice, tell their memories, and expressly authorize that this material be used after their death. This is the model followed by some of these services.
More problematic is the possibility of reconstructing someone once deceased without them ever having expressed whether they wanted to become an avatar. Technology allows this to be done using photographs, audio, or messages preserved by family members, but then the question arises of who can give permission to create that replica and what can be done with it.
What Spanish law says
Spain still does not have specific legislation on griefbots, although there are rules that can be applied. The Organic Law on Data Protection allows family members, related persons, and heirs to request access, rectification, or deletion of certain data of a deceased person, unless they have left express instructions to the contrary.
The Organic Law 1/1982 on honor, privacy, and one's own image also intervenes, allowing the protection of certain rights after death and establishing who can act in defense of the deceased. The regulation also considers certain uses of the name, voice, or image for advertising or commercial purposes as an unlawful intrusion.
Therefore, death does not automatically turn a person's voice, photographs, or identity into freely usable material to create a digital replica.
Spain prepares new rules for deepfakes and the deceased
The framework may soon be reinforced. The Government approved on July 7 the project for a new Organic Law on the civil protection of honor, privacy, and one's own image, currently in parliamentary processing.
The reform expressly addresses artificial intelligence and considers it illegitimate to use without authorization the voice or image of a person through AI for advertising or commercial purposes. It also aims to regulate the so-called digital will and allow a person to leave instructions regarding the future use of their image once deceased.
This may be especially important when technology is not limited to showing an old photograph, but allows generating phrases, videos, and conversations that the person never produced.
Europe requires identification of content created with AI
At the European level, some rules are already in force. As of August 2, 2026, the transparency obligations of Article 50 of the European Regulation on Artificial Intelligence will apply.
The system providers that generate or manipulate audio, images, videos, or texts must facilitate that these contents can be identified as artificially produced. The deepfakes are also subject to transparency obligations to warn that they have been created or altered by AI.
For an industry whose goal is to make a digital replica look like a real person, this obligation introduces an important boundary between reminding someone and making them believe that they are still really speaking.
Who decides when the avatar should disappear?
The questions do not end with the creation of these replicas. Researchers have warned of the possible effects of maintaining digital avatars of people in mourning for years and demand prior consent, clear warnings, and mechanisms that allow the definitive elimination of these representations.
It also remains to be resolved what happens if family members disagree about their use, who controls all the stored information, or to what extent a company can retain for decades the voice, memories, and intimate data of a deceased person.
The difference compared to a photograph or a traditional recording is profound. A file preserves something that a person really said or did; an artificial intelligence can use their face and voice to generate things they never said.
The question is no longer, therefore, whether it will be possible to talk with digital recreations of the dead. Technology already allows it. The debate is who can create them, what they can say on behalf of the deceased, and who will have the right to turn them off.