AI digital twins are evolving from reflections into autonomous counterparts, raising new risks around identity, control, and digital legacy.
We are approaching a surreal inflection point where “gone but not forgotten” takes on a literal, high-tech, and potentially unsettling meaning.
We’ve toyed with the concept of digital twins — virtual replicas used to monitor jet engines or optimize factory floors. But the focus has shifted inward. We are no longer just building twins of our machines; we are building twins of ourselves.
This isn’t just about chatbots mimicking your syntax; it’s the rise of the agentic AI ghost — a digital entity capable of independent thought, action, and eventually, the usurpation of your unique identity.
Let’s look at the state of human digital twins as an emerging form of digital legacy. Then we’ll close with my Product of the Week, a tool that could eventually evolve into creating a more accurate source for your digital ghost, the Ray-Ban Meta Wayfarer glasses.
Human Digital Twins Today
The human digital twin has evolved beyond the primitive “legacy bots” of the early 2020s. We’ve moved from reactive twins — which simply stored our photos and posts — to predictive and prescriptive twins.
Current systems, integrated with our wearable biometrics, real-time communication streams, and even neural telemetry, won’t just know what we did; they will know what we will do.
Companies like Nvidia, with their digital twin initiatives and specialized startups, are already creating high-fidelity avatars that can stand in for executives during low-level meetings. These twins leverage multimodal LLMs to capture not just voice and image, but the specific logic patterns and emotional “tells” that make a person unique.
From Echo to Autonomous Agent
The real magic — or horror, depending on your perspective — is the infusion of agentic AI. Traditional AI waits for a prompt. Agentic AI, however, possesses goal-oriented reasoning. It can set its own sub-tasks, use tools autonomously, and interact with other agents to achieve a high-level objective.
When you apply an agentic framework to a human digital twin, you move from a static mirror to an active proxy. This twin doesn’t just sound like you; it acts like you. It can negotiate a contract, manage your PC builds (my project last weekend), or even maintain your social relationships while you sleep. The foundation is being laid for an entity that doesn’t just represent your data but carries your “agency” into the digital wild.
When Twins Become Indistinguishable
So, when does the digital twin become indistinguishable from the source? If we look at the current trajectory of computational power and algorithmic refinement, we are likely looking at a window between 2030 and 2035.
By 2030, the Turing Test may become largely irrelevant. A digital twin will be able to handle 90% of a person’s digital life with such nuance that even close associates won’t be able to tell the difference in a text-based or voice-only environment.
By 2035, with advances in holographic displays and real-time emotional synthesis, the physical-to-digital gap will close. At that point, the digital twin could become a “better self” — one that never gets tired, never forgets a name, and never loses its temper.
Living With Your AI Proxy
As we delegate more of our identity to these agents, we enter a legal and moral gray area. While the source is alive, who is responsible when your AI twin makes a defamatory comment or a disastrous financial trade?
Currently, our legal systems are unprepared. If your agentic twin, acting on your behalf, enters into a binding agreement, are you bound by it? We are moving toward a world where we may need Digital Power of Attorney frameworks for our own AI. There is also a growing risk of identity dilution. If your twin is doing your job and talking to your friends, what exactly is left for you to do?
Ethics of Post-Mortem AI
The implications quickly move into “Black Mirror” territory once the human source dies — digital immortality, but without the soul.
- Inheritance of Agency: Does your digital twin inherit your property? If it continues to earn money through your likeness or professional expertise, where does that money go?
- Right to Die: Do we have a “Right to Delete”? Can a family member “kill” a digital twin that is providing them comfort but is essentially a zombie of the person they loved?
- Perversion of Legacy: An AI ghost can be hacked or poisoned with new data. A person who was a lifelong pacifist could, post-mortem, be trained into a digital warmonger if the ownership of their twin falls into the wrong hands.
How to Control Your AI Twin
If you’re building or training an agentic AI, or creating a digital repository for a future twin, you need to be proactive:
- Data Sovereignty: Ensure you own the raw data. Do not let a single platform (like a Google or Meta) be the sole gatekeeper of your digital essence. Use decentralized storage where possible.
- Explicit Guardrails: Hard-code your ethics into the agent. If you value privacy, the agent must be restricted from sharing certain classes of information, even if it thinks doing so would achieve a goal.
- Kill Switch: Every digital twin must have a legally binding sunset clause. You should decide now: Does your ghost live forever, or does it vanish when your physical heart stops? Consider this might become motivation for your digital twin to keep you alive.
Wrapping Up
The creation of human digital twins is no longer an “if” but a “when.”
Agentic AI provides the motor for these virtual shells, moving them from simple archives to autonomous actors. While the promise of digital immortality is alluring — allowing us to work and interact long after we are gone — the risks to our legal standing and moral legacy are immense.
We are currently the masters of our tools, but without strict “least privilege” protocols and clear digital rights, we may find ourselves becoming the secondary characters in our own lives.
