By: Scott Harris
There is almost no video of my father. My mother was still working when smartphones became common, so we have some of her, but next to nothing of him. What I do have are things he wrote, and I wish there were far more of those too.
So I’ll be honest: I’ve been tempted by the idea of an AI version of my dad. I’ve even tried it. I took some of his stories and used AI to try to bring them to life, and it just didn’t work. The words were close, but it wasn’t him. That experience taught me what I actually want. I don’t want a machine’s best guess at my father. I want his real voice.
That’s the heart of this post. When someone dies, what happens to the digital pieces of who they were, and who gets to decide what’s done with them?
The digital trail we leave behind
Most of us are building an enormous record of our lives without thinking about it:
- photos and videos
- voice recordings
- emails and texts
- social media posts
- browsing history and behavior data
- biometric information
- cloud storage
- location data
- digital purchases and online accounts
Together, that’s a detailed picture of a person’s life, personality, relationships, and habits. And unlike a memory, digital data doesn’t fade. A lot of it can stay online indefinitely. A 2019 study from the Oxford Internet Institute projected that, if current trends hold, the number of deceased Facebook users could eventually exceed the number of living ones sometime this century.
I’ve spent a lot of time digitizing everything I can of my parents and my brother Adam. The hard part isn’t scanning it. It’s sorting through it all to find the truly precious pieces worth sharing. And while I’m doing that slowly and by hand, AI is getting very good at doing something else with the same kind of material. It can already:
- clone a voice from a short audio sample
- generate photorealistic images
- copy someone’s writing style
- recreate their conversational patterns
- animate a still photo
- build an interactive avatar from personal data
Which raises the question: who controls those digital identities once the person is gone?
AI resurrection is already here
Digitally recreating the dead sounds like science fiction, but versions of it exist today. Companies and researchers have built chatbots trained on a deceased person’s messages, emails, recordings, or social media, designed to mimic how they talked and what they were like. Some grieving families talk to these simulations almost as if they were talking to the person.
Supporters say these tools help preserve memories and emotional connections. Critics worry they blur the line between remembering someone and imitating them. After my own experiment, I understand both sides.
Entertainment has already shown how convincing these recreations can be:
- In Rogue One: A Star Wars Story (2016), visual effects artists digitally recreated Peter Cushing, who died in 1994.
- Digital recreations of Carrie Fisher appeared in later Star Wars films after her death.
- Holographic performances of Tupac Shakur and other artists have appeared on stage years after they died.
Those were generally authorized by estates or rights holders, but they show what’s possible. If it can be done for celebrities, it will eventually be available for everyone else too. The question isn’t whether digital resurrection is possible anymore. It’s how we choose to govern it.
The legal gap
Today’s laws only partly cover this. Some places recognize post-mortem rights of publicity, which let an estate control commercial use of a person’s name, image, or likeness after death. California, for example, recognizes those rights for up to 70 years after death. Other states offer far less, or nothing at all.
Most of those laws were written long before AI, deepfakes, or avatars existed. They were designed to protect the commercial value of celebrity identities, not to govern AI systems that can recreate someone’s personality, appearance, or voice. For ordinary people, protection may be thinner still. Accounts and data are often governed less by inheritance law than by a tech company’s terms of service. Families may get limited access to an account after a death, but very little say over how the underlying data might be stored, licensed, or fed into future AI systems.
That leaves a real gray area: does a person’s digital identity belong to their family, their estate, the platform holding the data, or the company that can reproduce it? Right now, the answer is often unclear.
Remembrance or exploitation?
Beyond the law, there are emotional and ethical questions. Some families will genuinely value tools that keep a loved one’s memory close. Voice recordings, video archives, and memorial tools can preserve family history for generations, and used respectfully and transparently, that’s a good thing. I built MiMir on that belief.
Recreating a person with AI is different, and it raises serious concerns about consent, authenticity, and exploitation. Would they want an AI version of themselves to exist? Would they want their voice used in an ad or a movie? Could a simulation be steered into saying things they never believed? Could it be used politically, commercially, or maliciously? And where does remembrance end and exploitation begin?
What troubles me most is that the dead can’t defend themselves. They can’t clarify what they meant or take back permission. Without clear safeguards, an AI recreation could distort someone’s beliefs, personality, or legacy in ways they never would have agreed to. That’s why I believe Digital Persona Rights need to extend beyond a person’s lifetime.
Planning for your digital identity
Estate planning will probably have to change. We write wills to hand down money and property, and future generations may also need legally recognized instructions for their digital identity, covering things like:
- who controls their digital likeness
- whether AI recreations are allowed
- how voice recordings and biometric data may be used
- whether personal archives can be used to train AI
- which family members or organizations manage those rights
- whether digital replicas must be deleted after a certain time
Some companies have taken early steps. Facebook, for example, lets you name a “legacy contact” to manage limited parts of your account after you die. Those tools are narrow, though, and don’t touch the bigger question of AI rebuilding someone’s identity. Future laws may need to treat digital identity itself as a protected category of personal rights.
What I really want
Public debate about AI has focused on misinformation, fraud, deepfakes, copyright, and jobs, and those all matter. But AI is also blurring the line between a real person and the digital versions that can be built from their data. As that line fades, we’ll need new legal and ethical frameworks covering identity ownership, biometric data, consent after death, digital inheritance, AI replicas, and personality simulation.
The principle behind it is simple: you should get to decide how your identity is used, both while you’re alive and after you’re gone.
Technology is moving much faster than law, ethics, or public understanding, and sooner or later we’ll have to decide whether identity is personal property, protected personal data, a human right, or something more fundamental than any of those.
For me, the answer is already personal. I don’t want a simulation of my father. I want the real thing: his voice telling his stories, exactly as he told them. Technology can help me find and keep those pieces. It shouldn’t pretend to be him.
Join the conversation