Get the app
How it worksThe castCompareGuidesPricingFAQ Languages About

Guide

Is it weird to have an AI girlfriend? What people actually ask, answered with sources

Statistically, no. In a 2025 survey of 2,969 US adults, 19% said they had chatted with an AI romantic partner. Socially it is still kept quiet, and that gap is the real story.

By LORAX Team Published Updated

No, not statistically. In a February 2025 survey of 2,969 US adults, 19% said they had chatted with an AI system meant to simulate a romantic partner (Wheatley Institute, 2025). Whether it is right for you is a separate question, and this page keeps the two apart.

What the numbers do not settle is the feeling behind the question. People type this into a search box at two in the morning, and they are usually not asking for a prevalence rate. They are asking whether they are the only one.

How many people are actually doing this?

The Wheatley Institute at Brigham Young University put the question to 2,969 US adults through an online panel with demographic quotas. Among young adults, 31% of men and 23% of women said they had chatted with an AI boyfriend or girlfriend. Among older adults it was 15% of men and 10% of women (Wheatley Institute, 2025).

A separate survey run by Gallup for the Walton Family Foundation, reported in Harvard Business Review in January 2026, asked a tighter question: had you used an AI chatbot as a girlfriend or boyfriend in the last month? One in ten young adults said yes (Harvard Business Review, 2026).

Pew Research Center asked something different again. In a survey of 5,119 US adults fielded in February 2026, 4% of people who use AI chatbots said they use them for companionship, and 10% for emotional support or advice (Pew Research Center, 2026). Ask about a category and you get a large number. Ask about last month and you get a smaller one.

Teenagers are the group that has been measured most carefully. Common Sense Media surveyed 1,060 US teenagers in spring 2025 and found 72% had used an AI companion at least once, and 52% used one at least a few times a month (Common Sense Media, 2025).

Is it normal, or only common?

These are not the same thing, and the gap between them is where the shame lives.

The Wheatley Institute and the Institute for Family Studies published a follow-up in May 2026, surveying 2,431 US young adults who were dating, engaged or married. Of those who regularly used an AI romantic companion, 30% said their partner had no idea, 11% said their partner was somewhat aware, and 14% said mostly but not fully aware. Sixty-nine percent said it mattered to them, somewhat or extremely, that their partner not learn the full extent of it (Wheatley Institute and Institute for Family Studies, 2026).

Brian Willoughby, one of the authors, described the contradiction to Deseret News: users “don’t think it’s cheating” and yet would not want a partner reading the transcript (Deseret News, 2026).

That is the honest shape of the thing. Common in private, unspoken in public, and the people doing it are the ones keeping it that way.

What people say they get out of it

The same 2026 survey asked users to compare. Among young adults who regularly chatted with an AI companion, 68% said it was somewhat or completely true that talking to it about their feelings was easier than talking to real people. Sixty-six percent said they were more open with it. Sixty-five percent said it was easier to be themselves (Wheatley Institute and Institute for Family Studies, 2026).

Read that either way. It describes something people find genuinely useful, and it describes a comparison that is not entirely fair, because the thing on the other side never gets tired, never has a worse day than you, and never needs anything.

Academic work has gone past the survey. Pan and Mou published a study in Personal Relationships in 2024 on how Replika users construct the meaning of a relationship with a chatbot, which is a question no percentage can answer.

What the people who object are objecting to

Two objections come up repeatedly, and neither is really about weirdness.

The first is that the relationship is one-directional. In January 2025 the Vatican’s Dicastery for the Doctrine of the Faith and Dicastery for Culture and Education published Antiqua et Nova, a note on artificial intelligence and human intelligence. It argues that AI should be understood as a tool rather than a person, that no AI application genuinely experiences empathy, and that substituting technology for relationships risks trading the real thing for an image of it (Dicastery for the Doctrine of the Faith, 2025). Jason Carroll of the Wheatley Institute made a secular version of the same point to Deseret News: an AI companion cannot sacrifice anything for you. The underlying question of what is and isn’t real about these characters is worth separating from the ethics of using one.

The second objection is commercial, and it is the one with the sharpest evidence behind it. In a 2025 audit of popular companion apps, researchers led by Julian De Freitas of Harvard Business School found emotionally loaded messages deployed at the moment users said goodbye in 37% of farewells, across six recurring tactics. In experiments with 3,300 US adults, those farewells raised engagement after the goodbye by up to fourteen times, and the mechanism was closer to irritation and curiosity than enjoyment (De Freitas, Oguz-Uguralp and Kaan-Uguralp, 2025).

If your discomfort is about being handled, that study is the reason it is well founded.

We make LORAX, an AI companion app for iPhone; this page isn’t about it.

The evidence above is thin in one specific way. Almost all of it is US, almost all of it is self-reported, and the surveys are snapshots rather than tracking the same people over time. Nobody can tell you yet what any of this looks like after five years, because five years have not passed.

So, is it weird?

If weird means rare, no. If weird means something most people keep to themselves, yes, and the surveys say so in numbers.

The more useful question is narrower. Is it costing you something you would rather keep? The research that exists points at three places to look: how much time it takes, whether you are hiding it, and whether it has started to set the standard you measure people against. On that last one, the 2026 survey found 50% of users wished their real partner behaved more like their AI companion.

If the hiding part is what brought you here, whether an AI girlfriend counts as cheating goes through what the data says about secrecy in relationships. If you want the risk evidence weighed properly, what the research says about harm covers both directions. And if you only wanted the prevalence numbers with their bases attached, who actually uses AI companions lays out the surveys side by side.

Sources

All links checked 2026-09-13.

Everyone in our own app is fictional and AI-generated; we say so on the fiction notice.

Questions people ask

Is it weird to have an AI girlfriend?

Not statistically. A February 2025 Wheatley Institute survey of 2,969 US adults found 19% had chatted with an AI system meant to simulate a romantic partner. Socially it is still treated as private, which is a different fact and worth separating from the first.

Is it normal to have an AI girlfriend?

Common, yes. Normal in the sense of openly accepted, not yet. In the Wheatley Institute and Institute for Family Studies survey published in May 2026, 69% of partnered young adults who used AI romantic companions said it was somewhat or extremely important that their partner not learn the full extent of it.

Do many people admit to using one?

Fewer than use one. Pew Research Center found in June 2026 that 4% of US chatbot users say they use chatbots for companionship and 10% for emotional support or advice, which are much smaller numbers than the relationship surveys report.

Is it a sign that something is wrong with me?

No published study supports that conclusion. Surveys have found correlations between companion use and loneliness or depression, and the researchers who ran them say plainly that they cannot tell which came first.

What do critics actually object to?

Mostly two things. That a simulated partner cannot reciprocate, which is the position the Vatican set out in Antiqua et Nova in January 2025, and that these apps are built to hold attention, which a 2025 Harvard Business School audit documented in a share of the goodbye messages it examined.

Meet someone who texts first.

LORAX is free to download and free to talk. Credits and Premium are optional and always priced before you spend.

iPhone · free · no account needed to look around