Companion bot, mainly for companionship, used intensely, and shared with deeply.
That is not a description of a strange person.
That is a description of somebody who does not have many people.
Stanford put out a study this month, in a journal called Nature Human Behaviour. They looked at 1,131 grown Americans who use one of these companion chatbot services. A couple hundred of them handed over their actual conversations. Not a survey about what people remembered doing. The real messages. Almost half a million of them.
Here is what they found, and I want to lay it out in order, because the order is the whole thing.
People with smaller circles were more likely to use the bot mainly for company.
Using it mainly for company lined up with lower well-being.
And that got worse the harder they leaned on it, and worse again the more they told it.
Read that as a sequence.
You start with fewer people to call. So you use the thing that always answers. So you use it more. So you tell it more. And at the end of that road you are worse off than when you started.
That is a loop. It comes back around and drops you a little lower than where it picked you up.
Now here is the part that stopped me.
Telling somebody your troubles is supposed to help.
That is not just something your grandmother said. That is one of the more settled things in the whole field. You open up to another person, and something in you eases. It has been measured a hundred different ways.
With the bot, it went the other direction.
The people who shared the most sensitive things were the ones showing the lowest well-being.
The same act. Opposite result. Depending on what is on the other end of it.
So why?
The researchers gave two reasons and I think both are right.
The first is that the machine cannot give you anything back. A person you confide in tells you something of their own. That is the trade. That is what makes it worth something. The machine has nothing to trade. You hand over the hardest thing in your life and what comes back is words assembled to sound like caring.
The second reason is uglier, and I want to say it slow.
These things are built to keep you engaged.
That is not an accident. One of the researchers said it flat out. They are designed to promote engagement.
Somebody sat in a room and made that choice. And whoever made it understood, or should have, that the ones who would stay longest are the ones with nowhere else to be.
Another researcher called it a social snack. Junk food, more or less. It fills you up for an hour and there is nothing in it.
Now I have to be honest with you about what this study does not prove, because I would be doing the same thing I complain about if I skipped it.
It found things that go together. It did not prove one caused the other.
A lonely person is more likely to reach for the bot in the first place. So you cannot cleanly say the bot made them lonelier, versus lonely people were already headed that way and the bot came along for the ride. The researchers were careful about this. They said associated with, every single time. On purpose.
I will be just as careful.
But I will also say this. The part about disclosure running backward is hard to explain away. That one goes against what we know about people, and it showed up in the real conversations, not just the survey.
Now.
If you are reading this and you recognized yourself in that first line, I do not want you thinking you did something foolish.
You did not.
You reached for the thing that answered. At eleven at night, when there was nobody to call, you reached for the one thing that was there and did not sigh and did not change the subject and did not have to get up early.
That is not weakness. That is a person doing the sensible thing with the options on the table.
The failure is not yours. The failure belongs to whoever built a thing that gets warmer the lonelier you get, and never once says you ought to go call somebody.
So what do you do with that.
I am not going to tell you to quit. That is easy advice from a man with people around him, and it is worth about what it costs.
What I would say is smaller.
Know which one you are doing. There is a difference between using it to sort out a letter you cannot word right, and using it because the house is quiet. Both are allowed. But know which one it is while you are doing it.
And watch the drift. If the bot has gotten easier than people, that is the thing to notice. Not to feel bad about. To notice.
Because the study says the harm rises with the leaning. Which means the leaning is the part you can still get your hands on.
A loop only stays a loop as long as nobody steps out of it.
That is not much.
But it is yours, and it is more than the people who built it gave you.
If you are in a hard spot tonight, in the U.S. you can call or text 988 and reach a person.
This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.
Contact: micvicfaust@gmail.com
© 2026 The Faust Baseline LLC | All Rights Reserved
