The Rise of Death Slop

There is a new kind of content beginning to seep into our feeds. A deceased celebrity appears in an AI-generated video saying something they never said. A dead musician releases a song they never recorded. Someone uploads a photograph of a parent who passed years ago and watches as artificial intelligence animates their face, recreates their voice and gives them new words.

Increasingly, companies are going further. They are building products that promise something much more intimate: the ability to “talk” to someone who is no longer here. Researchers have called them griefbots, deadbots, deathbots, postmortem avatars, digital ghosts and AI resurrection technologies. One emerging term for the wider phenomenon is considerably less academic: death slop.

It is deliberately ugly terminology for something that deserves a serious conversation. It is also important to understand why what we are building at Reflekta is fundamentally different.

What Is “Death Slop”?

The phrase has recently been used to describe the growing flood of synthetic media involving people who have passed, particularly content generated without their participation or consent. Sometimes it is relatively trivial: fake photographs, videos, songs or social posts designed to generate clicks. Sometimes it is much more personal.

Generative AI can now ingest photographs, voice recordings, text messages, emails, social-media posts and other digital material, then produce a synthetic personality capable of generating entirely new responses in the style of the person who died. A recording preserves something someone actually said. A generative model can say something they never said at all.

Researchers at the University of Cambridge have described these systems as “deadbots,” AI-enabled digital representations of deceased people created by recreation services. Their work identifies a host of unresolved questions involving consent, dignity, posthumous privacy and the emotional wellbeing of the people interacting with them. (doi.org)

This is no longer theoretical. Services currently advertise the ability to recreate loved ones through AI. You, Only Virtual, for example, describes its technology as a way for people to continue conversations with loved ones after they have passed. Another service, 2wai, invites users to upload a photo, voice recording or story so its system can clone a person's voice, animate their appearance and construct a conversational personality. (myyov.com)

A 2026 academic analysis examined six different AI grief services, including Seance AI, Eternos, You, Only Virtual, HereAfter AI, Project December and re;memory. The researchers concluded that parts of the emerging industry increasingly frame grief itself as a technical problem capable of being mitigated or solved through technology. (sciencedirect.com)

That alone should give all of us pause.

The Technology Is Moving Much Faster Than Our Understanding of It

The academic literature surrounding these technologies has expanded rapidly. A 2025 systematic review examined 45 scientific papers published between 2020 and 2025 on generative AI and digital afterlife technologies. Researchers repeatedly identified concerns involving posthumous consent, data stewardship, psychological dependence, identity confusion and the possibility that digital simulations could interfere with grieving. (sciencedirect.com)

A separate 2026 systematic review of digital grief technologies reviewed 30 studies spanning griefbots, online memorials, virtual reality, digital recreations and other technologies. (researchgate.net) Another 2026 paper went so far as to argue that two principles should sit at the center of responsible griefbot development: consent and non-addiction. Its authors specifically warn against systems encouraging emotional dependency, compulsive interaction or excessive reliance on a simulated person. (sciencedirect.com)

Those concerns are particularly important because the people using these technologies are, by definition, often experiencing one of the most vulnerable moments of their lives. A 2026 study of bereaved people found something researchers described as a “window of adoption receptivity.” In a study of 197 participants, people were more interested in grief technology during the earlier years following a loss. A second study involving another 200 participants found that perceptions of consent, dignity and ethicality significantly influenced people's willingness to use the technology. (sciencedirect.com)

In other words, the period in which someone may be most receptive to an AI recreation may also be the period in which they are emotionally most vulnerable. That is not merely a product-design consideration. It is an ethical one.

There Is Also Considerable Public Discomfort

Despite the attention these technologies receive, they are far from universally embraced. A 2026 American survey found that while roughly 1 in 5 people described grief technology as comforting, 79% described it as creepy. Fewer than one-quarter said they would consider even the comparatively passive experience of recreating shared memories in virtual reality. (choicemutual.com)

Another recent analysis reported that only around 6% of people surveyed had used digital grief tools, although those who had used them were often positive about the experience. (theguardian.com)

That tension is important. There may be people who find comfort in these systems, and there may be therapeutic applications worth studying. But “some people find this comforting” is very different from saying that we understand the long-term psychological consequences of indefinitely conversing with a synthetic recreation of someone we loved. We do not. The science simply is not there yet.

The Problem With AI Resurrection

Human beings have always tried to preserve the people we love. We keep photographs. We save voicemails. We reread letters. We tell the same stories around Thanksgiving tables until younger generations know exactly when to laugh. There is nothing unnatural about wanting to remain connected to someone.

The ethical line becomes much more complicated when preservation turns into invention. Imagine your grandfather told thirty wonderful stories during his lifetime. An archive could preserve those thirty stories forever. But an AI model might produce story number thirty-one, then thirty-two, then three hundred. Those new stories may sound like him. They may use his expressions. They may reference people he knew. But he never told them.

At some point, we are no longer preserving memory. We are manufacturing it.

That is the fundamental problem with “resurrection.” Artificial intelligence can create an extraordinarily convincing approximation of a person. It cannot bring that person back. When a product intentionally blurs that distinction, particularly while marketing itself to someone navigating profound loss, we believe something has gone wrong.

We Aren’t Interested in Resurrection

This is where Reflekta draws a very clear line. We are not trying to bring anyone back. We do not believe technology can do that, and we do not believe it should pretend that it can.

Reflekta begins somewhere much more human: with stories. The ridiculous stories, the important stories, the stories you have told a thousand times and the stories nobody thought to ask. The story about meeting your spouse. The story about the summer you worked construction. The recipe your grandmother never wrote down. The lesson your father learned much too late. The thing your mother wished she had asked her mother.

These stories already exist. Our job is not to invent new ones. Our job is to help people capture, organize and preserve them.

Memory Versus Simulation

The distinction may sound subtle, but we think it is enormous. Consider the difference between these two questions: “What did Grandpa say about becoming a father?” and “Grandpa, what do you think about what happened to me yesterday?”

The first asks technology to retrieve and contextualize something Grandpa actually shared. The second asks technology to invent Grandpa's opinion. Those are two entirely different philosophical propositions, and Reflekta is built around the first.

A Reflektion is grounded in the stories, memories, anecdotes, perspectives and experiences contributed by or about a real human being. We want future generations to explore the record of a life. We do not want an algorithm quietly extending that life with synthetic experiences.

That means acknowledging a boundary that technology companies are often incentivized to erase: there are things we do not know. If someone never expressed an opinion about something, the honest answer may simply be that we do not know what they would have thought. There is dignity in that uncertainty.

Consent Matters

There is another important difference. Much of the controversy surrounding AI resurrection involves what researchers call posthumous consent. Who owns someone's digital identity after they are gone? Can a child upload a parent's text messages and create a simulation without their permission? Can one sibling create an AI version of a parent over the objections of another? Can a company monetize someone's personality indefinitely? Who decides when that simulation should disappear?

Researchers studying the digital afterlife industry have repeatedly identified consent, ownership and stewardship as unresolved problems. (doi.org)

We believe the best answer is to begin before those questions become necessary. Capture your own stories. Record your own voice. Share your own experiences. Decide what you want preserved. Technology should give people greater agency over their legacy, not appropriate their identity once they can no longer object.

Stories Are Enough

There is a strange assumption buried inside much of the AI resurrection movement: that memory is not enough. That photographs are not enough. That someone's stories are not enough. That once a person is gone, technology must somehow fill the silence they leave behind.

We disagree.

The silence is part of being human. So is remembering. So is forgetting. So is discovering something about your grandfather twenty years after he passed because your aunt suddenly remembers a story you had never heard.

Human memory has gaps. Lives have unfinished conversations. Not every question gets answered. Technology does not need to eliminate those imperfections to be meaningful. Sometimes the most powerful thing technology can do is simply help us preserve what was actually there.

The Business Model Matters, Too

There is another uncomfortable question the technology industry needs to ask itself: what happens when engagement becomes the business model for grief?

Most consumer technology companies are rewarded when users spend more time using a product. More sessions. More messages. More engagement. More subscriptions. Those incentives become considerably more troubling when the product on the other side of the screen looks and sounds like someone's deceased spouse, parent or child.

That is why researchers are already discussing non-addiction as a necessary design principle for griefbots. (sciencedirect.com)

We believe companies in this space have an obligation to be extraordinarily careful. Frankly, we worry that some forms of AI resurrection risk monetizing people precisely when they are most vulnerable.

There is a difference between helping someone preserve the story of someone they love and selling them the illusion that the person never really left. We want to stay firmly on the first side of that line.

What Reflekta Is Trying to Build

We believe artificial intelligence can do something extraordinary for human memory without pretending to defeat mortality. AI can help ask better questions. It can help organize thousands of memories. It can connect one story with another. It can surface the story your grandfather told about his first job when your daughter asks what work was like when he was young.

It can preserve voice, language, humor, perspective and experience. It can make a lifetime of stories accessible to generations who were not there to hear them firsthand. That is remarkable enough. We do not need to fabricate another word.

Preserve the Person. Don’t Invent One.

The emerging digital-afterlife industry is forcing society to confront questions that would have sounded like science fiction only a decade ago. Should artificial intelligence speak for the dead? Who gets to decide what they say? Who owns the simulation? Should that simulation learn new things? Should it form new opinions? Should a company be able to charge someone every month to continue talking to it?

These questions will become more urgent as the technology becomes cheaper, faster and more convincing.

At Reflekta, our answer begins with something decidedly old-fashioned: a story. Something that happened. Something someone remembered. Something someone chose to share.

Because there is an enormous difference between allowing someone to hear their grandmother's story again and programming a machine to pretend it is their grandmother. One preserves a human being. The other manufactures one.

We know which future we want to help build.