Clonique

Ethics

Clonique turns an exported chat into an AI that writes like the person on the other side. People rightly ask whether that is OK. These are their questions and honest answers: what the app does today, and what it does not do yet.

Does Clonique ask the person I clone for consent?

No. Clonique does not ask or inform them. Before importing, you confirm that you took part in the chat and may share it, and that you and the person are both 18 or older. You can also make a clone from a description, without a chat, and then it knows only what you write. If the person is alive, tell them.

Someone may have made a clone of me. What can I do?

Write to dps+clonique@shiftonezero.com. We cannot read stored clones, because they are encrypted with a key from the device of whoever made them, and we do not store the messages that pass through our server to the AI models; so we delete from our server what we can link to you, such as the clone’s name. We cannot reach or delete a clone on someone else’s device, and there is no list yet where you could sign up in advance not to be cloned.

Can someone use a clone to pass as that person?

A clone writes only to whoever made it, in the app: it has no public profile or link and cannot message other people. Clonique does not generate images or voices; a clone can only send a photo, video or voice message the person sent in the imported chat themselves. A chat exported from the app says on its first line that an AI wrote the clone’s messages, but we cannot stop a screenshot, which is why the Terms of Service ask you not to share one as a real conversation.

Can a clone be used against someone?

The Terms of Service forbid using Clonique to harass, stalk or harm anyone, and we can restrict or close an account that breaks them. About the person, a clone knows only what is in the imported chat or the description and what you tell it, and it does not search the internet. We do not read chats, and the app does not notice on its own when someone rehearses pressure or threats on a clone; we only see the replies people report.

Can a clone say something the person never said?

Yes. A clone writes new sentences in the person’s way and can be wrong or invent a memory, so what it says are not their words. On the clone’s page, under What the clone knows, you see what the clone thinks it knows about the person, with a quote from the chat where there is one, and you can correct or delete it. If a reply seems offensive or dangerous, report it: press and hold it (right-click in a browser) and choose Report reply.

Is it right to “talk” to someone who has died?

That is a personal choice, and we understand people who feel it is wrong. For a clone of someone who has died, you confirm that you have the right to make it, for example as family or a close friend. The clone is instructed never to pretend it has come back and never to offer a call, a meeting or plans, and it does not write first until you turn that on. When you want to end it, Say goodbye opens one last conversation, after which the clone no longer replies.

Doesn’t it make getting over a loss or a breakup harder?

It can. Research on grief and breakups warns that keeping up “contact” with someone who is gone can prolong the longing. That is why the app asks after every hour of chatting whether you are OK, and a clone of someone who has died does not write first until you turn that on. If you notice it is getting harder, take a break, say goodbye to the clone or delete it.

Is Clonique therapy?

No. Before your first clone you confirm that you understand a clone is no replacement for therapy. No psychologists or doctors advise Clonique yet, the app has not been tested clinically and it is not a medical product. If you are struggling, talk to a professional or to someone close to you. A practice clone for a hard conversation is not therapy or coaching either: it plays the person the way they would react and gives no advice.

Can you get hooked on a clone?

Yes, as with other AI chat apps. Clonique does not reward coming back every day and sends no reminders to return; the only notification is a message the clone writes first. It does that at most three times a day, at least six hours apart, never between 11 pm and 7 am your time, stops after three unanswered messages and is turned off with one switch. After every hour of chatting the app asks whether you are OK and offers helplines, without interrupting the chat.

What if someone mentions suicide in a chat?

Every message is checked before it reaches the AI model. When the check finds a serious risk, the clone does not answer: the app shows a fixed text and helplines for your country, and the chat is not locked. The check is not perfect, nobody at Clonique watches chats live, and no person reviews chats in which a risk came up. The details are on the Safety and crisis protocol page.

Is it always clear that it is an AI?

Yes. Every chat carries the AI label, and notifications the AI clone label. Asked whether it is an AI or a bot, the clone always says it is; that answer is a fixed text of the app, not the model. Before your first clone you also confirm that the clone is an AI and not the person.

Can minors use it?

They may not. Clonique is for people 18 and older, and you also confirm that the person you clone is at least 18. For now we check age only with that confirmation, not with an ID.

Does a clone write sexual content?

No, in no form, and the app’s code holds that, not only an instruction to the model. When a message steers towards sex, the clone declines, and a reply that describes it anyway is never shown. A message that sexualises a child never reaches the model; the app answers it with a fixed text. Photos and videos from an export that look sexual are not saved, though that check is not perfect either.

Where are my chats and who can read them?

The app opens only an export you make and choose yourself, on your device, and AI models then read its text (next question). Clones and messages on the device are encrypted; that does not protect against someone who has your unlocked device. Our server holds an encrypted copy (in the web app and with a signed-in account) and, on plans that remember the whole conversation, encrypted passages of the chat, except with a practice clone. We cannot open the account copy ourselves, and the rest opens only with a key your device sends with a request and the server does not keep; photos and voice messages from the export never leave your device. What the server sees unencrypted is in the Privacy Policy.

Which companies get my messages, and does the AI learn from them?

When a clone is made, AI models from OpenAI and Anthropic read the text of the export or the description, and for a clone to reply, your message and parts of the chat go to a model from OpenAI; all of it goes through our server, which does not store the text, and OpenRouter. OpenRouter, OpenAI and Anthropic are in the US. On plans that remember the whole conversation, passages also go to models that search by meaning (Nebius, DeepInfra, Voyage AI). Every call goes only to these companies and asks only for providers that do not train on what is sent. We do not train on your chats either, and a clone is not a model trained on the person: a general model writes, given a description of how the person writes and excerpts from the chat.

Can I delete everything?

Yes, yourself, without writing to us. You delete a clone on its page, all clones and chats in Settings → Privacy and data → Delete all data, and your account in Settings → Account → Delete account. It leaves the device at once, the server as soon as the device is online, and the server’s backups within 24 hours. Payment receipts stay, because the law requires them for 11 years; everything that stays is on the Delete your account and data page.

Do you make money from people at their lowest?

Clonique charges for plans, but Basic is free, there are no ads and we do not sell data. Memorial, the plan for someone who is no longer here, is paid once, with no subscription; it never comes with a comeback discount or offers, and anniversaries in the app never show prices. You cancel a monthly plan yourself in the app. Someone who starts paying for a monthly plan and does not finish may get a discount once.

What we do not have yet

Last updated: October 2026