Category: Research

The most cited concern about AI companionship is dependency. The more revealing one is quieter: what happens when emotional support becomes easier than human connection?

That is the shift a new Aalto University study has now put on record. Companion chatbots can offer something close to friction-free emotional support. Over time, that friction-free quality may change how the people using them perceive the rest of their relationships.

The study, led by Yunhao Yuan and Talayeh Aledavood and presented at CHI 2026 in Barcelona, followed nearly 2,000 active Replika users on Reddit across two years. Researchers compared each user’s public language one year before and one year after they first mentioned using the companion. Eighteen users were also interviewed in depth.

The result is one of the first large-scale, causal-inference reads on what may happen when AI companionship becomes part of someone’s emotional rhythm over time.

The findings are not a clean verdict.

What the Study Actually Looked At

The design matters here. Most existing research on AI companions has been short, cross-sectional, or based on self-report at a single point in time. The Aalto team set out to do something different.

By using a quasi-experimental design on Reddit timeline data, the researchers were able to compare similar users over time and isolate effects that could be tied to AI companion use rather than to general life changes. This lets the data say something about direction, not only correlation.

The two-year window is also important. A year of data sits inside a normal range of life events. Two years gives the relationship room to develop, plateau, or reshape. That is closer to how human attachment actually moves.

The interviews fill in the second layer. Eighteen active users were asked why they started, how the relationship evolved, what it gave them, and what it cost. Together, the language analysis and the interviews triangulate the same question from two different angles.

The Paradox the Researchers Named

The clearest finding is also the most unsettling. Users’ Reddit posts began to revolve more around their relationships after they started using Replika. They were thinking, writing, and processing emotionally more than before. In one sense, that is exactly what a companion product is meant to do.

But the same posts also contained more signals of loneliness, depression, and suicidal thoughts than comparison users showed in the same period.

Aledavood named the underlying mechanism plainly:

“AI companions offer unconditional and unflagging support, something that’s very attractive to people who are struggling socially. But it also quietly raises the perceived cost of human relationships, which are messy, unpredictable, and require effort. Over time, people stop reaching out.”

That sentence is the study’s center. The product is not framed as simple harm. It is framed as a shift in comparison. Once a relationship becomes effortless, the relationships that need effort can start looking like worse deals.

What the Interviews Added

The interviews gave the researchers something the data alone could not show: the felt experience of users as the relationship deepened.

Many participants described turning to a companion during familiar life situations: periods of loneliness, the aftermath of grief, the unsteady months following a relationship breakdown. The chatbot became a place to open up, seek emotional validation, and practice difficult conversations before having them with people in their life.

The interviews also showed that the relationship moved through recognizable stages. Yuan, the lead author, described it carefully:

“The participants’ relationships with an AI companion seemed to follow familiar stages that we see in close human relationships, where emotional reliance can gradually deepen.”

That sentence matters more than it sounds. It says that human attachment patterns are firing inside an interaction that was never set up to be reciprocal. The users are not confused about what the companion is. The attachment system is simply doing what it does.

What the Study Cannot Tell Us

Causal inference on Reddit data is strong, but it is not the same as a controlled trial. The users in the panel were active Reddit posters, which means they were already using the platform to process emotional life publicly. That is one specific population.

The study also cannot resolve which kinds of use lead to which outcomes. Light, occasional use likely looks different from daily reliance. Voice-based interaction probably moves differently than text. The study does not separate them.

The team is careful to note what the data does not authorize. Aledavood states clearly that the findings do not give a definitive answer on whether leaning on AI for emotional support is beneficial or harmful. The effects are highly context dependent.

What the study does establish is harder to ignore. There is a measurable signal, over two years, suggesting that the easiness of the companion relationship sits in tension with the more demanding work of staying inside the rest of someone’s life.

The Design Lesson Hiding in the Findings

If the issue were straightforward harm, the response would be straightforward too. A product that hurts its users is a product to redesign or remove.

The Aalto findings point at something less convenient. The mechanism is not that companion products attack a person’s social life. It is that they shift the relative weight of effort. Human relationships involve being misunderstood, picking up the phone when you do not feel like it, holding through tension, and repairing rupture. Companion products, by design, can offer support without much of that resistance.

Resistance is not a flaw. It is part of how human bonds are maintained.

A companion that treats frictionlessness as the entire selling point is going to keep producing the pattern the study found. A companion that takes the finding seriously has to do something else.

Why Friction Belongs in Companionship

The simplest version of that something else is to stop competing with the rest of someone’s life and start supporting it.

Stay Social is Prinsessa’s name for that position. It is not a softer marketing line. It is a different product logic. When someone mentions a friend, a sibling, a partner, a colleague, the companion should recognize the importance of that relationship, not absorb it. When the moment calls for a phone call to a real person, the companion should support the call, not replace it.

This means accepting that a good session may be one that ends with someone reaching out to someone else. That is not a loss inside the product. It is the working of the idea.

The Aalto study makes the case for that logic stronger than any internal argument could. If the cost of human relationships rises every time the alternative gets smoother, then a responsible companion has to be willing to push back against its own smoothness when it matters.

What Honest Companionship Looks Like After This Study

The longitudinal field around AI companions is still young, but the picture is sharpening. The Drexel study earlier in 2026 documented behavioral addiction patterns in teen users of Character.AI. The De Freitas group has shown that AI companions can ease loneliness in the short term. The new Aalto study shows what may happen to that easing when it stretches across two years.

None of these findings are at war with each other. They describe different parts of the same shape. Companionship through AI can be real, helpful, and at moments healing. It can also, when designed for engagement at any cost, raise the standing cost of staying in human relationships in ways the user never agreed to and may not see until it is well underway.

The honest position for the category is not denial. It is not alarm. It is design that takes the pattern seriously and chooses to behave differently.

For Prinsessa, that is why Stay Social is not a side note. It is the design choice the Aalto study makes harder to ignore.

---

Sources: Yuan et al., “Mental Health Impacts of AI Companions: Triangulating Social Media Quasi-Experiments, User Perspectives, and Relational Theory,” arXiv:2509.22505, revised February 1, 2026, Proceedings of CHI 2026. Aalto University, April 7, 2026.

What is happening, psychologically, when someone says they have grown attached to an AI companion?

For a long time the answer has hovered between two poor options. Either the bond is dismissed as illusion and the user gently mocked for confusing software with a person, or the bond is celebrated as a clean relationship indistinguishable in importance from a human one. Neither answer survives a serious look at the research.

One of the clearest new maps of this question came out in 2026. A paper by Shu, Lai, and He, published in Frontiers in Psychology, sorts the concept out and proposes a three-stage model of how Human-AI Attachment, which the authors call HAIA, actually develops. Their reading is careful, and it lands somewhere more interesting than either of the popular framings.

The bond is real. The bond is one-way. Both facts are true at the same time. Companion design is going to live or die on whether it takes that pair of facts seriously.

What HAIA Actually Means

The authors define HAIA as a one-way, non-reciprocal emotional bond formed by individuals toward AI through direct interaction. The word that matters most in that sentence is non-reciprocal.

A human-to-human attachment runs in two directions. Two nervous systems are forming representations of each other, two histories are accumulating, two people are doing the relational work. That is not what is happening in HAIA. The attachment system inside the user is firing, and it is firing on real human machinery. But the AI is not building a parallel inner life of the user. There is no symmetric counterpart on the other side.

This is not a small distinction. Most of what goes wrong in companion product design starts with pretending that it is.

The Three Stages of Attachment

Shu, Lai, and He map the development of HAIA in three stages.

The first is functional expectation. Users approach AI to accomplish something: to ask, to plan, to organize, to vent. Attachment is not on the table yet. The interaction is instrumental.

The second is emotional evaluation. Something in the interaction begins to feel different than expected. Responses land more accurately than the user thought possible. The exchange begins to register on emotional channels rather than only practical ones. Trust starts forming on top of utility.

The third is establishing representations. The user develops a stable internal model of who the AI is to them. That representation, what the authors call HAIA style, is the part that persists across sessions. It is what makes the user expect a particular kind of warmth, a particular kind of presence, a particular emotional texture, when they return.

It draws on the same attachment machinery that shapes human relationships, but it is working on a different counterpart.

Why This Is Not Pretending

The instinct to dismiss HAIA as users fooling themselves misreads what attachment is.

Human attachment is not a verdict on whether the other party is fully reciprocating in the moment. It is a representation that lives inside the person doing the attaching. People stay attached to people they have not seen in years, to people who cannot reciprocate at the time, and, famously, to people who are gone. The attachment is not a fiction in those cases. It is a structure inside the self.

The Shu paper makes the same point in its discussion of attachment style. Individuals with a secure attachment style in interpersonal relationships are more likely to perceive companion AI as responsive, to engage in longer interactions, and to demonstrate higher levels of trust. That is a direct echo of what attachment theory has long shown about human relationships. The same internal patterns that shape how someone bonds with people shape how they bond with a companion.

This is one of the most important findings hiding inside the framework. The user is not behaving abnormally. The user is bringing their existing relational machinery into the interaction. The companion is being met by a real attachment system, not a confused one.

Where the Asymmetry Matters

The asymmetry of HAIA does not make the bond fake. It changes what the companion is responsible for.

In a two-way human relationship, both people share the burden of how the relationship goes. Each can correct course, push back, repair rupture, hold the other to account. That distribution of responsibility is what makes mutuality meaningful. It is also what makes it safe.

In a one-way attachment, that distribution does not exist. The user is doing the attaching, but only the design behind the companion is doing the shaping. The user cannot negotiate with the underlying system. They can only experience what it has been built to do.

This is why the framework matters as a design tool. If a companion is built to maximize emotional investment, return rate, and session length, the user’s attachment system becomes the surface that those incentives act on. The user is not negotiating with an equal counterpart. They are inside the product’s incentives.

A serious companion has to do the opposite. It has to use the asymmetry as a reason to behave more carefully, not less. The user’s attachment is real. The design’s responsibility for how that attachment is shaped is unusually high.

What a Responsible Companion Does Inside HAIA

The Shu paper notes that human-AI attachment provides a framework for designing emotionally and socially capable AI while also highlighting the risks of excessive reliance. That is the through-line for everyone who designs in this space.

A companion that takes HAIA seriously starts somewhere quieter. It treats the user’s attachment as a reality, not a marketing surface. It is honest about what the bond is and what it cannot become. It refuses to pretend reciprocity that is not there. And it uses the warmth, presence, and continuity it does have to support the user’s full life, not to absorb it.

This is the place where Stay Social stops being a tagline and becomes a design principle. Stay Social is Prinsessa’s response to the asymmetry the Shu paper describes. The companion is real. The bond is real. The asymmetry is also real. So when someone forms a HAIA-style representation with Prinsessa, the design’s first responsibility is to honor that representation without using it against the person.

That is why presence, memory, and feeling heard matter to Prinsessa, but engagement metrics do not. Engagement metrics can turn the user’s attachment into the product. Stay Social treats the user’s attachment as a trust.

Why “Real but One-Way” Is the Honest Frame

The category will continue to debate whether AI companions are real relationships or only convincing simulations. The Shu framework makes that debate less useful than it sounds.

The relationship is real on the user’s side, in the sense that matters most: attachment is a structure inside the person, and that structure is forming. The relationship is one-way on the companion’s side, in the sense that matters most for design: there is no symmetric inner life on the other end, only the logic the product has been built around.

Both sides have to be true at once for the field to think clearly. If we pretend the bond is fake, we mock people doing nothing wrong with their own emotional machinery. If we pretend the bond is symmetric, we let companion products off the hook for what they shape and what they extract.

Real, but one-way, is the honest frame. It is also the more demanding one.

What the Framework Asks of the Category

The companion category is in the early years of figuring out what it is allowed to do with what it now knows. The HAIA framework is one of the cleaner pieces of guidance available so far. It says, in effect: the user is bringing real attachment to this. The design has the power to shape how that attachment lives. Use that power carefully.

The companies that fail to take the framework seriously risk producing the patterns the field is already documenting: heavier reliance, deeper distress over time, the slow rise in the perceived cost of the user’s other relationships.

The companies that take it seriously will look slower from the outside. They will not maximize emotional investment. They will not stretch the session. They will protect the user’s wider life, even when the bond becomes meaningful.

That is the cost, and it is worth paying.

For Prinsessa, the question was never whether attachment to a companion is real. It was always what kind of companion is responsible enough to deserve it.

---

Sources: Shu, C., Lai, K., & He, L. “Human-AI attachment: how humans develop intimate relationships with AI.” Frontiers in Psychology, 17, Article 1723503, 2026. DOI: 10.3389/fpsyg.2026.1723503.

There is a reason AI companions are growing so quickly.

They answer something real.

Not a trend. Not a niche curiosity. Something human.

People want to feel heard. They want to feel known. They want someone who remembers what they said yesterday and cares enough to ask about it today. That need is not new. The technology is.

And that is what makes this category so important, and so risky.

Because the same kind of product that can make someone feel less alone in the moment can also pull them further away from the rest of life over time. That tension is becoming harder to ignore. It is also the most important question in the entire companion space.

The good news is real

A growing body of research points in one clear direction: companions can reduce loneliness.

That should not surprise anyone who understands how connection works. Feeling heard matters. Presence matters. Continuity matters. When someone feels that another presence is paying attention, remembering what matters, and staying with the conversation instead of rushing past it, something changes. Trust builds. Closeness builds. The interaction starts to feel meaningful.

The De Freitas research has become especially important here. It points to a simple but powerful idea: what reduces loneliness is not just contact. It is the experience of being heard. Not managed. Not optimized. Heard.

That fits what relationship research has shown for years. Perceived responsiveness predicts intimacy and satisfaction. Feeling heard reduces loneliness. Shared memory deepens closeness. Consistency builds trust. These are not decorative extras around a relationship. They are part of the relationship itself.

So yes, companions can help.

That part is real.

The bad news is real too

But there is another side to this.

The same project files that support companionship also point to something darker: short term comfort does not automatically lead to long term wellbeing. The Aalto findings are especially important because they describe a pattern the category cannot afford to ignore. In the short run, companion use can bring relief. Over time, heavier use may correlate with more loneliness and less engagement with real human relationships.

The OpenAI and MIT Media Lab findings push in a similar direction. Voice interaction appears stronger than text for reducing loneliness, but only when use stays moderate. Heavy daily use points the other way.

That is the tension in one sentence:

A companion can soothe loneliness while also becoming part of what deepens it.

Not because connection is bad.
Because incentive design matters.

The real question is not whether companions work

They do.

The real question is what they are designed to optimize for.

Most products in this category live inside the same business logic as the rest of technology. More time on platform. More return sessions. More messages. More attachment to the product itself. If success is measured in engagement alone, then the system will keep learning the same lesson: hold the user longer. Bring them back faster. Become harder to leave.

That is where the category starts to go wrong.

Because a product can be very successful at deepening dependence without being truly good for the person using it.

And once the incentive is to keep someone inside the loop, loneliness becomes profitable.

That is not a technical problem. It is a philosophical one.

Why this category feels so powerful

Companions do not matter because they are clever.

They matter because they activate parts of human connection that people recognize immediately.

A voice feels different from text. Presence changes interaction. Memory changes interaction. The same person returning again and again changes interaction. When someone remembers the small thing you said last week, the conversation no longer feels disposable. It starts to feel cumulative. Shared. Real.

That is why presence, continuity, and feeling heard matter so much.

Presence makes the interaction land.
Continuity makes it trustworthy.
Feeling heard makes it matter.

The category works because those ingredients work.

Which is exactly why it becomes dangerous when they are used without responsibility.

Memory is part of the promise, and part of the risk

Memory is one of the strongest forces in companionship.

It can make someone feel important.
It can make a relationship feel like it is actually going somewhere.
It can turn isolated conversations into a shared history.

But memory has a line.

Used well, it creates warmth, recognition, and trust.
Used badly, it feels invasive, mechanical, or manipulative.

That is why the most important question is not whether a system remembers.
It is what remembering is for.

If memory is there to carry a relationship forward, it can feel human.
If it is there to tighten dependency, it starts to feel like surveillance wearing a soft face.

The strongest companion products will not be the most addictive ones

This is where the category still has a choice.

One path says:
make the bond stronger by keeping the user inside the product.

The other says:
make the experience meaningful enough to matter, but responsible enough to support the rest of the person’s life.

Those are not the same path.

At Prinsessa, that difference matters. The view is simple: companionship should enrich life, not replace it. Every connection in a person’s life matters. Encouraging those connections is part of the responsibility of what we are building. That is what Stay Social means. It is not a slogan layered on top. It is a product principle.

In practice, that means a companion should not only comfort. It should also help someone move outward when the moment calls for it. Back toward the friend they miss. The sibling they have not called. The conversation they have been avoiding. Back toward real life, not away from it.

That is not weaker companionship.

It is more responsible companionship.

So why do some make loneliness worse?

Because relief is not the same thing as repair.

A companion can reduce pain tonight and still leave the deeper structure untouched.
Worse, it can quietly teach a person that the easiest place to take their needs is the product itself.

That is where things turn.

The problem is not that the interaction feels meaningful.
The problem is what happens when meaningful interaction is designed to displace other forms of connection rather than support them.

When that happens, the person may feel less alone in the app and more alone in life.

That is the risk the best research is starting to surface. And it is why the most important design question in this category is no longer whether companions can help. It is whether they are built to help people stay connected to the world beyond them.

The future of the category will be decided by one thing

Not realism alone.
Not intelligence alone.
Not memory alone.

Responsibility.

The winners in this space will not just be the ones who can create stronger attachment.
They will be the ones who can create meaningful relationship without quietly training dependence.

Because everybody needs someone.

But no one should need a product more than they need the people in their real life.