Do People Really Use AI Girlfriends? The Data

Updated 2026-06-10. AI girlfriend guides

Yes, clearly. This is not a speculative category, and the skepticism it still attracts is out of date. But the numbers attached to that yes are mostly not real.

This page previously opened with "approximately 33 million monthly active users globally," a market "exceeding 1.2 billion dollars," a user base that "more than tripled since early 2023," and a breakdown showing 75 percent male users with an average age of 27. None of those figures can be sourced. There is no authoritative count of AI companion users, no audited market value, and no industry-wide demographic survey. The category has no regulator collecting data, no standard definition of what counts as a user, and platforms that treat their metrics as competitive intelligence.

We would rather tell you what is genuinely knowable than publish a table of invented numbers with confident sourcing.

Correction, October 5, 2026. Every statistic previously on this page — the user count, message volume, session length, retention rates, gender split, age distribution, regional breakdown, motivation percentages, and the claim that fMRI studies showed AI conversations activate the same neural regions as human bonding — has been removed. The last of these is worth calling out specifically: it is a precise neuroscientific claim about brain imaging, and we could find no study supporting it.

What is actually established

Adoption is real and large enough to matter. AI companions are used at scale, consistently across app stores, web traffic, and payment processors. Replika has reported millions of users, and several products in this category have crossed into mainstream visibility. That much is observable even without a good industry-wide count.

The user base skews young and male. This is the one demographic pattern with reasonable support, though the precision is not. The widely observed pattern is that early adoption skews male and under-35, and that the gap narrows over time — but nobody has published a defensible breakdown by exact percentage, and the figures that circulate (typically "75 percent male, average age 27") trace to unattributed industry blogs repeating each other.

Motivation is mixed, not singular. Loneliness is frequently cited as a driver, alongside curiosity about the technology, creative writing, roleplay, and social skill practice. Surveys of this kind exist but are small and self-selected, and the categories overlap heavily — someone who starts out curious often stays for different reasons than they arrived.

Emotional attachment is documented. Users do form attachments, and this is measurable. The clearest evidence is a survey of 612 long-term users of AI companion apps in mainland China, which found usage frequency correlated strongly with emotional attachment (Liu, Frontiers in Psychology, 2025).

What nobody knows, and why

How many people use these products. Counts are compiled by different firms using different methods — app download estimates, web analytics, payment data — and they do not agree. Add-avoid web usage is invisible to all of them, and people with more than one app are counted more than once.

How much money the category makes. Revenue estimates exist and vary by an order of magnitude depending on what gets counted.

The retention curve. Thirty-day and ninety-day retention figures are quoted constantly and sourced to nobody. Plausible shapes exist — a steep first-week drop, then a long tail of engaged users is what every consumer subscription looks like — but the actual levels are not public.

How much this substitutes for human contact. This is the question that matters most and the one we can least answer at scale. The best available evidence is small: a preregistered 21-day randomized trial of 183 participants found no significant difference in social health or loneliness between people chatting with a companion and people playing word games (arXiv 2509.19515), and a two-wave study of 1,182 Character.AI users found sustained engagement predicted lower wellbeing through reduced in-person interaction with others (arXiv 2609.07243).

How you can find out for yourself

If you are trying to answer "how common is this," your own data is better than the industry's. Look at app store download rankings in your category, which are audited and public. Check whether any major platform has published figures in a filing or interview — a few have, and those are the only numbers worth quoting. And notice that most of the loudest statistics come from companies selling the category.

Why we are publishing a page with fewer numbers

An honest version of this page ranks worse for search terms like "how many people use AI girlfriends." We would rather have the page that survives someone actually checking it. If you are doing research on AI companions, the fabricated figures are more likely to mislead you than to help you — and citing one puts your own work at risk.

Frequently Asked Questions

How many people actually use AI girlfriends?

There is no reliable public count. Several individual companies have reported their own user numbers, and those are the most trustworthy figures available. Industry-wide totals circulating online are estimates built on stacked assumptions, and they disagree with each other.

Do people really form emotional bonds with AI?

Yes — this is the least speculative part of the whole picture, and it is measurable. A survey of 612 long-term users found usage frequency correlated strongly with emotional attachment. What we do not have is evidence that this bond is comparable to human attachment in any meaningful sense, and we removed a claim about shared neural activation from this page because no study supports it.

Is AI girlfriend use growing or declining?

It is growing, and app store rankings and web traffic consistently point that way. Precise growth rates are not public and any percentage you see attributed to a specific period is an estimate.

How long do sessions last?

We do not publish our own session data, and no credible industry-wide figure is available. What you can check is your own screen time, which is more informative about whether your use is going the way you intend.

What percentage of users are female?

Nobody has published a defensible figure. The qualitative pattern is that early adoption skewed heavily male and has been narrowing, with the fastest recent growth among women — but the exact percentages that circulate are unsourced.

This article is for informational purposes only and does not constitute medical or professional advice. Luvlock AI provides tools for companionship but encourages maintaining a balanced approach to relationships and mental health.

Ready to experience AI companionship for yourself? Get started with Luvlock AI →