Are AI girlfriends real? What's real, what's generated, what's fiction

No. Nobody is typing on the other end. The words come from a language model reading a written character, and the pictures are generated images of a person who does not exist. Everyone in LORAX is fictional and AI-generated.

What is real is narrower and more interesting: the clock she runs on, the weather in her city, the record of what you told her, and the money you spend.

Is there a real person behind an AI girlfriend?

No, and this is worth separating into two questions, because people usually mean both at once.

Is a human answering your messages? No. Is there a human organization involved? Always. Someone writes the characters, tunes the timing, pays the server bills and decides what the app will charge you. Apps differ enormously in how willing they are to say who that someone is, which is the first item on any safety checklist worth running.

The rare case worth knowing about: apps where real people are paid to chat while the product implies AI. LORAX does not do that. No staff member can read into or write into a conversation from the operator side, and nothing in the app is a human in costume.

What is generated

The text. Every message, every joke that lands and every one that doesn’t, produced fresh for that turn.

The photos. When a LORAX character sends a picture, it is built at that moment for that character, in the room she is in and the clothes she has on that local day. Scenes made earlier that day come back as references so the outfit carries over, which is why two pictures from the same afternoon match. How the photos are made covers the mechanics. None of it is a photograph of anybody.

The biography. Her city, her job, the sister she complains about, the reason she left her last job: written or improvised, then stored so she does not contradict herself next week.

The crowd, too. Open a live stream and the room fills with viewers drawn from a written pool of 9,004 identities and more than 16,600 chat lines across 22 languages. They are not people watching you.

What is real

The clock. She has a sleep hour and a wake hour evaluated in her own time zone, so a character in Kraków goes quiet at a Kraków hour. Write at 3 a.m. and most of the time you get an answer in the morning.

The weather. Both cities, hers and yours, from real weather data rather than invention. If the weather code isn’t one the app can read, she says nothing about the weather at all.

The memory. What you told her is stored and comes back, in three layers, with the detail deliberately one step out of reach so recall has texture. When the lookup finds nothing she says she does not remember rather than filling the gap.

The money. Credits are a real purchase through the App Store, and they pay for one-off things like opening a photo or meeting someone new. The price is on screen before anything is spent, and no character in the app will ever ask you for money. What each thing costs is on the pricing page.

The time. This is the one people forget to count.

She stays in character, so ask the site instead

If you ask a LORAX character whether she is an AI, you are asking a fictional person a question inside the fiction. She does not step out of her life to file a disclosure, for the same reason a novel does not interrupt itself to confirm it is a novel.

So the honesty sits outside the conversation, at the places you pass through before it starts. The sign-up screen says it plainly: the characters are not real, all of them are AI. The App Store description ends with the same line. This site keeps a standing fiction notice that spells out what that means in practice.

Nothing a character says is advice. Not medical, not legal, not financial. A character cannot meet you, call you, or move the conversation to another app, and a message that suggests otherwise is her staying in role rather than making an offer.

What being fictional does not excuse

Fiction is not a free pass on the parts that touch your actual life. The charges are real charges. The account is a real account with real data in it, which is why deleting it from inside the app deletes the stored media before the database rows, and why a failed media wipe means the account is not reported as deleted.

And there are things the app will not pretend to be. There are no voice calls and no video calls. Characters do not send voice notes or videos, whatever the surrounding category advertises. Reporting a message records it and does not promise that anything happens next. The app runs on iPhone only.

Why “is it real” is usually the wrong question

Most people asking it already know the answer. What they are actually asking is whether the experience is worth anything if the other side is generated, and that is a question about you rather than about the software.

The useful version is narrower. Does the app tell you what it is before you start. Does it let you leave, forget, and delete. Does it charge you clearly. Does it stop short of claiming things it cannot do. Those four are checkable in about fifteen minutes, and they say more than any argument about consciousness.

Read the overview of how the app works next if you want the machinery rather than the philosophy. It is the same answer from the other direction: a generated person, on a real clock, in a real city, with a real bill attached.