What a text clone actually captures
A text clone is built from what you have written: chats, posts and notes. From that material, a language model can pick up a surprising amount:
- Style: sentence length, punctuation, emoji habits, how you open and close messages.
- Phrases: pet names, running jokes, the words you overuse.
- Tone by relationship: people write differently to a parent, a partner and a colleague.
- Stored memories: dated facts and events that appear in the messages, which can be retrieved when relevant.
- Opinions and attitudes, to a degree. In a 2024 Stanford-led study, AI agents built from two-hour interviews with 1,052 people matched their survey answers 85% as accurately as the people matched their own answers two weeks later.
That last result is impressive, but notice what it measures: answers to survey questions. Reproducing what someone would say is not the same as being someone.
What a clone lacks
| A clone can have | A clone does not have |
|---|---|
| Your writing style | Your experiences, feelings or sensations |
| Facts that appear in its data | Memories of anything after its data ends |
| Plausible replies in your voice | Intentions, goals or wishes of its own |
| Patterns from many conversations | A body, a daily life or relationships of its own |
Consciousness. A 2023 report by 19 researchers, led by Patrick Butlin and Robert Long, assessed AI systems against leading scientific theories of consciousness and concluded that no current AI systems are conscious. A clone produces text about feelings without having them.
Agency. A clone does nothing until it is asked, or until a scheduled feature triggers it. It does not want to talk to anyone.
A life of its own. You keep changing after the export. The clone does not. It also cannot know what you did yesterday unless you add new material, and it can invent details that sound right but are false.
Is an AI clone a person in law?
AI systems are not legal persons under the law of the EU, the UK or the US, and a clone cannot own property, sign a contract, vote or be held responsible. The people and companies who build and use it carry that responsibility. Laws differ by country; this is general information, not legal advice.
The idea has been discussed. In 2017 the European Parliament asked the European Commission to consider, in the long run, a special status of “electronic persons” for the most sophisticated autonomous robots. A large group of AI, robotics, law and ethics experts objected in an open letter. The EU’s AI Act, adopted in 2024, regulates AI systems and the duties of those who provide and use them, without making AI a legal person.
Courts have drawn similar lines. In March 2025 a US federal appeals court, in Thaler v. Perlmutter, confirmed that US copyright law requires a human author, so an AI system cannot be listed as one.
Your rights over your own likeness
Although the clone has no rights, the person it imitates does. Several areas of law can apply:
- Data protection. In the EU, messages about a living person are personal data under the GDPR.
- Personality and publicity rights. Tennessee’s ELVIS Act, signed in March 2024, added protection for a person’s voice, including simulated voices. California’s AB 1836, signed in September 2024, requires consent from the estate before a digital replica of a deceased performer’s voice or likeness is used in audiovisual works or sound recordings.
- New proposals. Denmark has proposed amending its copyright law so that people control realistic digital imitations of their appearance and voice. A European Parliament briefing in January 2026 expected it to take effect in mid-2026.
- Transparency. From August 2026 the EU AI Act requires chatbots to make clear that users are talking to an AI, unless it is obvious, and requires realistic “deep fakes” to be disclosed.
Consent: whose clone is it?
A clone of yourself, built from your own messages, is the simplest case. Even then, your chats contain other people’s words, so treat them with care.
Cloning another living person is different. Their style and memories belong to their identity, and they should agree first. Memory Clone requires that permission in its Acceptable Use Policy. For someone who has died, think about what they would have wanted and how close family feel. Our ethics checklist before cloning someone walks through these questions step by step.
How to talk about a clone honestly
Words shape expectations, especially for grieving people and children. Simple, accurate language helps everyone.
| Instead of | Try |
|---|---|
| “This is me.” | “This is an AI simulation of how I write.” |
| “Dad says…” | “The simulation answers the way Dad might have.” |
| “She lives on in the app.” | “The app keeps her style of writing.” |
| “It remembers everything.” | “It can find memories that appear in our messages.” |
- Tell people when they are chatting with a clone rather than with you.
- Never let a clone make promises, decisions or agreements in your name.
- Treat surprising statements as guesses, and check them against the original messages.
Where Memory Clone fits
Memory Clone is built around this modest view. It is a Windows app, currently in alpha, that creates a private AI simulation from chats you export yourself and notes you write. It learns style separately for each relationship and is designed to keep private facts from one relationship out of conversations with others. Imported originals stay on your PC, and every clone is presented as an AI simulation, not as the person.
If you want to see what a clone of your own writing feels like, the download page explains how to begin. Our plain-English definition of an AI clone is a good next read.
Frequently asked questions
Can an AI clone think like me?
It can produce answers that resemble yours, based on patterns in your messages. It does not think, feel or understand in the way you do.
Does an AI clone have rights?
No. AI systems are not treated as legal persons. The rights belong to the people involved, such as the person being imitated.
Will my clone know what happened after I made it?
Only if you add new material. Otherwise it reflects the period covered by its data and may guess about anything later.
Is it legal to make an AI clone of someone else?
It depends on the country and on how the clone is used. Getting the person’s clear permission is the safe and respectful starting point.
Could a clone ever become conscious?
Researchers disagree about the future. A 2023 expert report found no current AI system is conscious, and a clone built from messages is not designed to be.
Sources
Checked on . Third-party products and help pages change often; follow the links for current details.
- Park et al., “Generative Agent Simulations of 1,000 People” (2024, first version) (arXiv)
- Simulating Human Behavior with AI Agents (policy brief, 2025) (Stanford HAI)
- Butlin, Long et al., “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (2023) (arXiv)
- Open Letter to the European Commission on Artificial Intelligence and Robotics (2018) (Robotics Open Letter)
- Regulation (EU) 2024/1689 (Artificial Intelligence Act) (EUR-Lex)
- D.C. Circuit upholds human authorship requirement in Thaler v. Perlmutter (2025) (IPWatchdog)
- Tennessee’s ELVIS Act expands publicity rights for individuals’ voices (Manatt)
- California expands its post-mortem right of publicity law to cover AI digital replicas (Cowan, DeBaets, Abrahams & Sheppard)
- The Danish approach to copyright and deepfakes: A model for the EU? (2026) (European Parliamentary Research Service)
This article is general information, not medical, legal or financial advice. Memory Clone creates an AI simulation from the material you provide; it is not the person.