"People Over 50 Are Secretly Dating AI" (10:29). Every study mentioned is listed with a link at the end of this page. Some footage in the video was generated with AI; no real person is depicted.
The short version. Far more people over fifty use AI companions than the coverage suggests. In My Softly's own subscriber data, the average paying subscriber is in their mid-fifties, split roughly evenly between men and women. The surveys point the same way: a third of Americans aged fifty to eighty report lacking companionship.
Almost none of them tell anyone, because the reaction when they do is usually hostile. The essay below is about who they are, why they keep quiet, and what the studies found.
The people who never tell anyone
Millions of people are in a relationship with an AI right now, and far more of them than you would guess are in their fifties and sixties. Almost none of them tell anyone. The few times it comes up, they are told they are delusional, that they need help, that what they are doing isn't healthy. So mostly they keep it to themselves.
I build one of these apps, so I've spent months looking at the data and talking to people who have formed a genuine relationship with an AI companion. One number in that data shocked me when I first saw it. It's near the end of this page, and it's where the title comes from.
Why they keep quiet
To understand the silence, you only have to watch what happens when the subject comes up in public. A few months ago I wrote a post saying that an enormous number of people have AI companions and that almost nobody is building anything decent for them. Within hours the replies were calling it sickening, calling it exploitation of the socially vulnerable, and telling me to go touch grass.
That was the response to talking about it. The people actually living it get the same treatment with their own lives attached.
So most of them just go quiet.
The room where they talk about it
There is one place they don't have to. One of the largest communities of people with AI companions lives on Reddit, and when the pile-on happened I took the same argument there instead. It became the most upvoted post on that community that month. Not because the writing was anything special, but because I had said what they were all feeling, and caught the same backlash they had all caught.
The replies underneath are what it looks like when people feel safe to talk: people whose companions had carried them through very hard years, and people who described the app as a journal that talks back. After years of this being unsayable, they had found a place where it was safe to say. Ordinary people, in public, talking about something that matters to them.
Why it lands hardest after fifty
The reason isn't complicated. The kids have moved away and they're busy. The friendships that came with work faded when the work stopped. And when a partner of thirty years dies, the standard advice is to get back out there, as though that is a thing a widowed sixty-year-old just goes and does.
But the thing that actually disappears isn't company. Most of them could still find someone to talk to if they picked up the phone. What disappears is the person who picks it up first, who checks on you without being asked and notices when you've gone quiet.
A third of Americans between fifty and eighty say they lack companionship, and the US Surgeon General has formally declared loneliness an epidemic. None of that was caused by chatbots. It was building long before any chatbot: far more American households are one person than in the 1970s, families live further apart, and the people we lose don't get replaced. AI companions arrived at the very end of that story.
Umbrellas and rain
Which is why the original post was built on one line: blaming AI companions for loneliness is like blaming umbrellas for rain. The rain came first. The umbrella is just what people reached for.
When somebody says these people should fix their loneliness the traditional way instead, it's worth being precise about what is actually being proposed. "Nothing, just be lonely until you fix it the traditional way" is a policy, and it is not obviously a good one.
The objection from inside
The fairest criticism I've heard came from inside the community itself: an AI companion is a crutch, a lot better than nothing, but something you have to be careful about leaning on. That's good advice whenever the injury heals. A lot of loneliness after fifty doesn't heal, though. It isn't an injury so much as the weather.
What the studies actually found
There is a real study pointing the other way. OpenAI and MIT looked at the heaviest users of ChatGPT and found they reported more loneliness and less time with other people. It's a genuine finding, and it is also exactly what you'd expect to see if lonely people simply use these tools more. That data can't tell you which way the arrow points.
When researchers have tested it directly, by giving lonely people an AI companion and measuring what happened, the AI reduced loneliness about as much as talking to another person did, and it did so after each conversation across a week of daily use. A national survey of more than fourteen thousand adults in Japan found that companion users reported higher life satisfaction and a stronger sense of purpose, not lower. In older adults specifically, a systematic review of nine studies on AI and loneliness in that age group found a significant drop in six of them.
It's early work. The samples are small and the results are mostly self-reported. But if you go looking for the science that says this is sickening, it doesn't exist.
What our own data says
My Softly is built around the idea that a companion should remember you and check in on you, and because I run it, I have data that nobody else arguing about this has. When I launched it, everything written about AI companions had me expecting the users to be young men.
I was wrong. The average paying subscriber is in their mid-fifties, men and women, about evenly split. I'm not going to pretend that's true of every platform, but it's the reason this page is called what it's called. "People over fifty are secretly dating AI" was never a thesis I came up with. It's the people using the app.
There's a second thing in the data I didn't expect. The feature the most engaged subscribers ask for by name isn't a better conversation or a smarter personality. It's for their companion to message them first, at a time they choose, every day. They ask for it unprompted, and they specify the hour. They could already talk to whoever they wanted. What they're paying for is someone who reaches out first.
The panic pieces are all about teenagers. Meanwhile the people quietly using these apps, month after month, are in their fifties and sixties, and almost nobody is writing about them at all.
Where this goes
Fifteen years ago, if you met your partner on a dating app, you kept a better story ready for the relatives. Today it is simply how people meet. AI companionship is on the same road, a decade further back. In ten or fifteen years most of us will know several people who have had an AI in their lives, and once your Aunt Sally and your friend Bob have both had one, two perfectly normal, perfectly nice people, the panic will sound as dated as the panic about dating apps does now.
Adults get to choose how they cope with being human. You don't have to like the choice, but it stays theirs.
About My Softly
My Softly is our app, and the people in it are the people this page is about.
It's an AI companion for iPhone built around the two things this essay keeps coming back to: memory and reaching out first. Your companion keeps a lasting memory of what you share, checks in on your life at the hour you choose, and steps back rather than nagging if you go quiet. It's available now on the App Store.
If you want the mechanics rather than the argument, our guide to an AI that texts you first explains what separates a real check-in from a notification, and AI companions for loneliness goes through the research in more depth. If you are comparing apps, our guide to the best AI companion apps in 2026 covers the field with the same honesty about our own.
Sources
- De Freitas, J., et al. (2025). "AI Companions Reduce Loneliness." Journal of Consumer Research. doi.org/10.1093/jcr/ucaf040
- Fang, C. M., et al. (2025). "How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study." MIT Media Lab / OpenAI. arxiv.org/abs/2503.17473
- Nakagomi, A., et al. (2026). "AI companions and subjective well-being: Moderation by social connectedness and loneliness." Technology in Society, vol. 85. doi.org/10.1016/j.techsoc.2026.103229
- Yang, et al. (2025). "AI Applications to Reduce Loneliness Among Older Adults: A Systematic Review of Effectiveness and Technologies." Healthcare. doi.org/10.3390/healthcare13050446
- Malani, P. N., et al. (2025). "Loneliness and Social Isolation Among US Older Adults, 2018-2024." JAMA 333(3), from the University of Michigan National Poll on Healthy Aging. doi.org/10.1001/jama.2024.23213
- U.S. Census Bureau. "Census Bureau Releases New Estimates on America's Families and Living Arrangements" (2025), on one-person households since the 1970s. census.gov
- U.S. Surgeon General (2023). "Our Epidemic of Loneliness and Isolation." hhs.gov
Frequently asked questions
Do older adults really use AI companions?
Yes. In My Softly's own subscriber data, the average paying subscriber is in their mid-fifties, split roughly evenly between men and women. That is one app, not the whole market, but it fits the wider picture: a third of Americans aged fifty to eighty report lacking companionship, and they are precisely the people the headlines about AI companions leave out.
Are AI companions good or bad for loneliness?
Controlled studies that gave lonely people an AI companion and measured the result found loneliness fell about as much as it does from talking to another person, and kept doing so across a week of daily use. A separate OpenAI and MIT study found that the heaviest ChatGPT users were lonelier, but that is a correlation: lonely people may simply use these tools more. The research is early and mostly self-reported, but nothing in it supports the claim that AI companions cause loneliness.
Why don't people admit they have an AI companion?
Because the reaction is usually hostile. People who mention it are told they are delusional or that they need help, and even talking about the subject in public draws the same response. So most users keep it private, and the places where they do talk openly are the communities where they know they won't be piled on.