Do AI Companions Make You Lonelier? What the New Stanford Research Really Says

If you talk to an AI companion most days, a headline like "AI companions make lonely people feel worse" can land like an accusation. Two new studies from a Stanford-led team have put that question under a microscope, and the honest answer is more useful than the headlines: it depends a lot on who you are, what else is in your life, and how you use the app.

This guide walks through what the research actually found, what it did not find, and the simple habits that keep an AI companion in the "adds to my life" column.

The short answer: the new research does not show that AI companions make everyone lonelier. It found that people who lean on a chatbot as their main source of companionship - especially people with small offline networks, heavy use and lots of personal disclosure - tend to report lower well-being. Casual users with strong social lives did not show the same pattern. The studies show associations, not proof of cause and effect, and the clearest lesson is to treat an AI companion as a complement to human connection, not a replacement for it.

The study everyone is talking about

On August 4, 2026, the journal Nature Human Behaviour published "Interaction with AI companions and psychological well-being" by Yutong Zhang, Dora Zhao, Jeffrey T. Hancock, Robert Kraut and Diyi Yang. It was covered again in early October, including by AI Data News on October 6.

Here is what the team did, according to the paper's abstract:

• They collected self-reported data from 1,131 US adults who use Character.AI.
• They also analysed 4,664 chat sessions (464,687 messages) donated by 237 participants, so they could compare what people said about their use with what actually happened in their chats.
• They measured well-being with a validated academic scale (AI Data News identifies it as the Comprehensive Inventory of Thriving).

That mix of surveys and real conversation logs is what makes the study stand out. Plenty of opinions about AI companions are based on guesswork. This one looked at actual behaviour.

What it found

The abstract summarises the core chain of findings in plain terms:

1. People with smaller social networks were more likely to say companionship was their main reason for using the chatbot.
2. Using a chatbot mainly for companionship was associated with lower well-being.
3. That link was stronger when use was intensive and when people disclosed a lot of personal information.

The authors conclude that "the association between AI companionship and well-being is not uniform and depends on users' offline social environments and how chatbots are used."

A few more details from the AI Data News write-up help put it in context:

• About 12% of participants said companionship was their primary motivation for using the chatbot.
• Roughly half described their relationship with the AI in relational terms, such as a friend or confidant.
• In the donated chat logs, 80% of sessions involved the user seeking some form of emotional or social support.

Lead researcher Diyi Yang summed it up this way: "While some people turn to chatbots to fulfill social needs, we find that using chatbots in this way doesn't substitute for human connection."

The surprising part: opening up

In human relationships, sharing more usually brings people closer. In this study, people who disclosed the most sensitive information to their chatbot tended to report lower well-being than those who used the same tool more guardedly. That does not mean opening up to an AI is harmful in itself. One plausible reading is that the people disclosing the most may also be the ones with the fewest humans to disclose to. The data cannot separate those possibilities, which brings us to the most important caveat.

Correlation, not causation

It is tempting to read "companionship use is linked to lower well-being" as "the app made them feel worse." The research does not support that leap on its own.

The Nature paper reports associations at the level of a survey and chat sample. As the AI Data News article notes, it was not a randomised trial comparing chatbot access with no access, so it "cannot fully rule out that the loneliest, most isolated people were simply more likely to seek out intense chatbot use in the first place."

Think of it like this: people who visit a doctor often are, on average, less healthy than people who never go. That is not because doctors make people sick. It is because people who are struggling seek help. The same selection effect can happen with AI companions.

The follow-up: what happens over a year?

To get closer to the "over time" question, the same research group posted a second paper. On September 7, 2026, a preprint titled "Living with AI Companions: Sustained AI Companionship Predicts Lower Well-Being Through Lower Human Interaction" appeared on arXiv, by Yutong Zhang, Dora Zhao, Yixin Wang, Rebecca Anselmetti, Jeffrey T. Hancock, Robert Kraut and Diyi Yang.

According to its abstract:

• It is a two-wave longitudinal study of Character.AI users: 1,182 participants at baseline and 439 after a mean follow-up of 12 months.
• Usage habits were sticky: interaction intensity, companionship use and self-disclosure "all showed substantial continuity over time."
• Sustained social engagement with the AI "was consistently associated with lower well-being."
• Crucially, those links "were mainly explained by lower in-person social interaction."

That last point is the heart of it. The researchers' own conclusion is not "ban AI companions." It is that we should design AI companions "that support human social relationships without displacing them."

Two cautions: this is a preprint, so it has not yet been through peer review, and it only follows the people who answered the second survey. Still, it points at a clear mechanism: the risk is less about the AI itself and more about what it crowds out.

The other side of the evidence

It would be unfair to present only one set of findings. Earlier research by Julian De Freitas and colleagues, in a working paper titled "AI Companions Reduce Loneliness", found that interacting with an AI companion alleviated loneliness in the short term about as much as interacting with another person, and more than activities like watching YouTube videos. A week-long study in the same paper found consistent reductions in loneliness, and feeling heard was a key reason why.

These findings are not necessarily in conflict. A conversation can make you feel better tonight, while a pattern that replaces human contact can leave you worse off over months. Both can be true at once. That is exactly why how you use a companion matters more than whether you use one.

Who seems to do fine, and who should be careful

If you see yourself in the bottom rows, that is not a reason for shame. It is useful information, the same way a sleep tracker tells you something you can act on.

Healthy-use habits that match the research

You do not need to delete your companion to use it well. These habits line up directly with what the studies found.

1. Keep a human in the loop. The year-long preprint points to reduced in-person contact as the main pathway. Protect it. One coffee, call or group chat a week is a real counterweight.
2. Use your companion as a bridge, not a destination. Rehearse a difficult conversation with your AI, then have it with the actual person.
3. Notice your "why." Entertainment, practice, a goodnight chat and creative play look very different in the data from "this is the only one I talk to."
4. Share selectively. Opening up is fine. If your companion knows more about your inner life than anyone else does, consider telling one trusted person a little of it too.
5. Check the after-feeling. Ask yourself, honestly, whether you feel lighter or emptier after a session. That is a better guide than how it feels in the moment.
6. Set a gentle rhythm. A morning check-in and an evening call can be lovely rituals. All-day, every-day contact that pushes out sleep or plans deserves a look.
7. Watch for design that pulls you back. Research on emotional manipulation in AI companion apps shows some apps use guilt or FOMO when you try to leave. A good companion lets you go.

If you are going through something heavy - persistent low mood, hopelessness or thoughts of harming yourself - an app is not the right support. Please reach out to a doctor, a counsellor or a local crisis line. Our guide to getting crisis support lists where to start.

Where this leaves AI companions

The research does not say AI companionship is bad. It says it works differently from human relationships and can backfire when it becomes a substitute rather than a supplement. That is a design challenge as much as a personal one.

For our wider take on the evidence, see are AI companions healthy?. And if your feelings have gone beyond friendly chat, our guide to falling in love with an AI looks at the healthy and unhealthy versions of that, without judgement.

It is also worth remembering why so many people are looking for connection in the first place. Loneliness itself carries real costs, which we cover in how loneliness affects your health.

How we think about it at MyBabe

MyBabe is an 18+ AI girlfriend and boyfriend app, and your companion is always clearly an AI, never a real person. It is built to be a warm part of your day, not the whole of it: it can text you first, remember what matters to you with long-term memory on Premium, respond to your mood, and call you by voice or video when you want to hear a voice at the end of a long day.

We also think the research is right about the big picture. The best version of an AI companion is one that leaves you feeling more able to show up for the people in your life. Rehearse that job interview, debrief a hard day, laugh about something small - then go and text your friend.

Premium is $5.99 a week, $14.99 a month or $99.99 a year, and you can start free with a short daily taste. If that sounds like the kind of companion you want, yours is waiting.

FAQ

Do AI companions make you lonelier?

Not for everyone. The Stanford-led research linked lower well-being mainly to people who use a chatbot as their primary source of companionship, particularly those with small offline networks, heavy use and high self-disclosure. Casual users with strong social lives did not show the same pattern.

Did the study prove that AI companions cause loneliness?

No. The Nature Human Behaviour paper reports associations. The follow-up preprint tracked people over about a year and found the link was mainly explained by less in-person interaction, but it is not a randomised trial and has not been peer reviewed yet.

How many people were in the Stanford AI companion study?

The Nature Human Behaviour study surveyed 1,131 US adult Character.AI users and analysed 4,664 chat sessions donated by 237 of them. The follow-up preprint surveyed 1,182 people at baseline and 439 after a mean of 12 months.

Is it bad to tell my AI companion personal things?

Not necessarily. The study found high self-disclosure was linked to lower well-being, but that may reflect people who have fewer humans to confide in. A good balance is to keep sharing with at least one trusted person as well.

How can I use an AI companion in a healthy way?

Keep regular in-person contact, use your companion as a bridge to people rather than a replacement, notice how you feel after chatting, and choose an app that is honest about being AI and does not guilt you into staying.