Digital afterlife law protects your estate. Not your consent

Somewhere in the fine print of your estate plan, sandwiched between the house and the retirement accounts, there is now a question that did not exist five years ago: what happens to your digital self when you die? Not your passwords. Not your social media accounts. Your voice, your face, your decades of messages — the raw material from which an AI system could be trained to continue speaking as you, answering questions, responding to your children in the night, perhaps indefinitely.

The United States Senate has an answer. The NO FAKES Act, advanced by the Senate Judiciary Committee with bipartisan support in June 2026, establishes a federal digital replica right that extends seventy years past a person’s death. It gives estates the legal standing to block unauthorized AI reconstructions of a deceased person’s voice and face. It is, in the field of posthumous identity law, genuinely significant. It is also built on an assumption that deserves much closer scrutiny than it has received: that controlling your digital afterlife is a matter of property law, and that property, at death, passes to your heirs.

The law has finally arrived at the problem. It has handed the answer to the wrong people.

The technology that made this necessary

For most of the last decade, digital afterlife technologies operated in a legal and ethical near-vacuum. Companies including HereAfter AI — co-founded by journalist James Vlahos as a voice-interview preservation tool — and StoryFile, which pioneered recorded-video interaction with the deceased, built their products around a simple proposition: that families want to preserve a loved one’s voice and memories, and that technology could help. The demand was genuine. The ethical scaffolding was not.

What the industry discovered, the hard way, is that preservation is the easy part. The harder question is whether the person being preserved consented to the specific uses their data would be put to. A grandmother who recorded her life stories for her grandchildren did not necessarily agree to serve as a training corpus for an AI that her heirs could license to grief counselors. The distinction matters enormously, and for most of the industry’s existence, no one enforced it.

The technologies now available go further than preservation. Deadbots — a term that has entered academic literature — are AI systems fine-tuned on a deceased person’s communications: their texts, emails, social media posts, voice recordings, personal journals. A 2026 study published in Frontiers in Digital Health, drawing on interviews with twenty-five participants, found that griefbots are perceived as neither inherently beneficial nor harmful. Users reported emotional comfort alongside risks: disruption of natural grieving processes, perceived inauthenticity, and dependency on an entity frozen at the moment of the last data point.

The psychologist Julia Shaw has put the deeper problem plainly: AI is a perfect false memory machine. A deadbot that says something plausibly-but-inaccurately characteristic of the person it simulates does not just fail to represent them — it actively reshapes how the living remember who that person was.

What the NO FAKES Act actually does

The NO FAKES Act is a serious piece of legislation. Tennessee’s ELVIS Act, signed in 2024, was the first US statute to address AI posthumous voice and likeness use directly. The NO FAKES Act federalizes and extends that protection. What it establishes is a post-mortem intellectual property right, modeled on copyright: the seventy-year window mirrors the copyright term, the right descends to the estate, and unauthorized AI recreations can be blocked or litigated.

This closes the legal vacuum that allowed companies to train systems on dead people’s WhatsApp archives without asking anyone. It gives families a cause of action when their father’s voice is fed into an advertising campaign he would have found mortifying. These are real harms, and the law addresses them.

But the model it borrows — intellectual property, held by the estate, transferable and licensable — carries a structural assumption that the debate has largely not interrogated: that a person’s digital identity is analogous to a novel they wrote, or a song they recorded. That after death, the question of who gets to use it is the same kind of question as who gets the house.

The case for leaving it with the family

The counterargument runs something like this: the dead cannot be harmed. They cannot be embarrassed, distressed, or manipulated. The people who can be harmed — by an unauthorized recreation, by a malicious deepfake, by a grief technology that exploits their loss — are the living. And the people with the most direct stake in how a deceased person’s memory is managed are, in most cases, the people who loved them. Defaulting to family control, in the absence of any instruction, is the least-bad option among a set of imperfect choices.

There is also the practical argument. Requiring every adult to complete a detailed post-mortem data directive while alive — to specify, in advance, what may and may not be done with their voice, their face, their message archives — is a bureaucratic demand that most people will not fulfill. Most people do not have a will. They certainly do not have a digital afterlife policy. The NO FAKES Act, on this reading, is the best achievable law in a world where most people have not thought about this at all.

Why property is the wrong category

The problem is that this argument concedes the philosophical point too easily. Consider what the medical field did with a structurally identical problem. Advance directives — living wills, do-not-resuscitate orders — exist precisely because we recognized that decisions about bodily intervention after incapacitation are not property decisions. They are personhood decisions. The right locus of authority is the person themselves, expressed while they were capable of expression. Next-of-kin authority is a fallback, not a principle.

Researchers at the University of Cambridge have proposed exactly this framework for digital identity: the digital do-not-reanimate order, or DDNR. Like a DNR, it would allow any competent adult to specify whether they consent to posthumous AI simulation — in what forms, for what purposes, with what restrictions. France has already moved in this direction: French data protection law allows individuals to define directives concerning the retention, deletion, and communication of their personal data after death. The right runs with the person, not the estate.

What we know, and what is genuinely contested

The technology to create convincing posthumous simulations exists and keeps improving. Demand for such technologies from grieving families is real and not going away. The legal vacuum until the NO FAKES Act was genuine and produced real harms. The seventy-year estate right solves for commercial exploitation by strangers. It does not solve for a family member who uses a deceased person’s data in a way that contradicts that person’s values.

The organ donation analogy is imperfect but instructive. Organ donation registries resolved a version of this problem by making the individual’s expressed preference legally determinative. A national DDNR registry, connected to estate planning infrastructure and digital platform account settings, would do the same for posthumous AI use. It would require the legal system to acknowledge something the NO FAKES Act stops just short of saying: that your digital self is not a financial asset, and that the right to determine its fate belongs to you first.

The law has arrived. The question now is whether it builds from the right premise.

Tags: ai, Digital afterlife, NO FAKES Act, digital identities, griefbots, post-mortem data rights