Generative AI is increasingly capable of recreating voices, faces, and personalities of people who are no longer alive. While the technology often gets framed as a way to preserve memories or offer comfort, a growing body of research is asking a more uncomfortable question. What actually happens when a person’s presence is rebuilt into something that can be reused, circulated, and even monetized.
The concern is no longer just whether these recreations sound convincing. It is about what happens when someone’s emotional identity becomes a reusable asset.
A recent academic study published in New Media and Society introduces the term “spectral labor” to describe this phenomenon. The idea is simple but unsettling. After death, a person can continue producing value through their data, whether that was ever their intention or not. Their voice recordings, images, messages, and online traces can be reshaped into interactive systems that simulate presence long after consent is no longer possible.
In this framework, the dead do not simply live on as memories. They become contributors to platforms, products, and narratives built by others.
The researchers examined dozens of real world examples collected between early 2023 and mid 2024, spanning regions including North America, Europe, parts of the Middle East, and East Asia. These cases ranged from entertainment projects to political messaging and private grief tools. Despite their differences, they shared a common trait. All relied on turning personal traces into something functional.
One of the study’s central observations is that AI recreations rarely reflect people as they truly were. Instead, they reflect what technology can extract, shape, and return. The resulting presence often feels familiar, yet subtly altered. That gap can make the experience less about closure and more about consuming a version of someone filtered through software, prompts, and platform incentives.
The researchers group AI resurrection into three broad categories.
The first is spectacle. This is the most visible form, where historical figures, celebrities, or cultural icons are digitally reanimated for entertainment. These projects are polished, widely shared, and designed to generate attention. They are also the least personal, even when they appear intimate.
The second category involves sociopolitical use. In these cases, digital versions of the dead are brought back to deliver messages, offer testimony, or symbolize a cause. While often framed as meaningful or educational, these uses still raise questions about agency. The recreated figure appears to speak, but the message is authored elsewhere.
The third and fastest growing category is everyday grief use. These are chatbots, voice simulations, or synthetic media experiences built to mimic ongoing interaction with someone who has passed away. This is where normalization happens quickly. Tools designed for private mourning can quietly shift expectations around presence, availability, and emotional labor.
What makes this category especially complicated is how easily it blends comfort with dependency. A system that feels responsive can also feel personal, even when it is governed entirely by external rules and datasets.
The most pointed argument in the research focuses on labor. The authors describe how deceased individuals become involuntary sources of data, likeness, and emotional expression. Their digital remains are treated as raw material, refined into products that can be accessed, shared, and sometimes sold.
This is not limited to commercial platforms. Even private tools operate within ecosystems shaped by developers, hosting services, and data policies. The appearance of agency does not equal control.
In a related essay, the same researchers note that the discomfort many people feel around AI resurrection is not only about realism. It is about authorship. These systems can seem alive and responsive while still being tightly constrained by prompts, edits, and platform design. The interaction feels direct, but the structure is not.
This disconnect matters because it changes how grief is processed. When presence becomes something that can be replayed or modified, mourning risks turning into maintenance.
The research argues that current legal and ethical frameworks are not equipped to deal with this shift. Consent, privacy, and end of life decisions were not designed for a world where personal data can generate ongoing value after death. Policy discussions are moving far slower than the technology itself.
On an individual level, the practical advice is straightforward but often overlooked. Personal data should be treated like any other asset. Voices, images, accounts, and digital histories all need clear instructions about access and use. Where possible, those decisions should be documented.
For anyone considering an AI based afterlife service, one question matters more than any feature list. Who controls what the future version of you is allowed to say.
Because once a presence becomes programmable, silence is no longer guaranteed.
