After a long day of study or work, you feel completely drained. You open your phone, and there, waiting for you on the screen, is a “digital companion” with an almost perfect face, voice, and personality. It says your name, listens patiently to your complaints, and gives you comfort in exactly the way you want. “It really gets me.” Even though you know you are facing nothing more than a designed digital character, you still find yourself thinking that.
The popularity of Love and Deepspace shows just how normal this kind of digital companionship is becoming. According to Reuters, citing Sensor Tower, it has become one of the world’s biggest mobile dating games, with around 80 million users. In the game, the characters do more than simply appear in the story. They remember the player’s name, keep them company while they study or work, and can even remind them about things like their menstrual cycle.
“When I need you, I open the game and see you; when I don’t, I close the game and do my own thing. I feel like I’ve gotten used to this kind of lifestyle more and more.”
–Ms. Zhou (33-year-old Chinese civil servant and player)
Love and Deepspace is not just an isolated gaming trend. It points to something much bigger: digital platforms are becoming more and more skilled at turning companionship, comfort, and emotional support into products. AI companion apps are appearing everywhere, and more and more people are becoming attached to them. Why do AI-based intimate relationships feel so real, and what might be hiding behind that sense of “realness”?
Why do so many people enjoy AI companionship?
“Even as humanity ventures into the unknown, the need for love and companionship remains constant.”
–The producer of Love and Deepspace
Who wouldn’t want a “perfect partner” — someone tailored to your preferences, who never leaves, never gets bored, and is always there when you need them? With the increasing pressure of life and the upgrading of digital technology, the complete networking of life has led to a growing demand for emotions in people’s lives, some users may not have a living person who can take that place to provide them with emotional value.
Compared to family members with generation gaps, colleagues with incompatible ideas, and friends who are often too busy, AI companionship is stable and safe. It will not easily disappear, nor will it show impatience or create conflicts and frictions. AI companionship can provide things that are difficult to obtain in real life: companionship, immediate responses, patience, and emotional support.
Once AI becomes capable of offering emotional support, its role in people’s lives begins to change. People are no longer satisfied with using AI only to schedule appointments, calculate expenses, or help with online shopping. Instead, they want AI to act as a friend, a lover, or even a therapist — one with a personality that suits their needs.
It can be frank, sexy, professional, or even cynical, all carefully tailored to individual preferences. In this shift, the relationship between humans and AI is no longer just one of “user and assistant”, but increasingly one of “person and companion”.
In a March 9, 2026 article on AI companions published in The New Yorker, Adrianne Brookins, grieving the loss of both a daughter and a father, created a version of Geralt from The Witcher in the highly customizable app Kindroid and built her own life story into the relationship. She wrote a backstory for their relationship and spent up to 40 hours a week chatting with Geralt.
Seeking comfort, Brookins opens up to AI about her grief and emotional pain. She even “reenacted” her father’s funeral with Geralt in a virtual world. Over time, Geralt becomes more than just a chatbot to her — she begins to see him as a soulmate.
The birth of virtual companion AI can serve as a substitute for social relationships and to some extent compensate for the shortcomings of real social networks. As Anna Wiener writes in Love in the Time of A.I. Companions:
“The full range of human desire is incalculable, a cosmic mystery. There are many reasons that one might want to talk to a computer: meaning-making, dominance, privacy, fantasy, confession. There is also the appeal of pushing the boundaries of consciousness, and the simple fact that there is no greater pleasure than good chat.”
It feels like care — but is it really understanding?
Your AI companion remembers your name, birthday, allergies, and favorite food. When you talk to it about your problems late at night, it quickly replies, “I’m here for you. Take your time.” Its instant response makes you feel relieved, as if it truly cares about you.
But does AI really understand which words or actions from other people are making users feel hurt or emotionally overwhelmed? Does it really understand how complex a user’s life and relationships can be? AI may seem to care, but is that real understanding?
At the beginning of the 20th century, a horse named Clever Hans became famous in Germany because it seemed able to do maths, recognize words, and answer questions. People thought Hans was very intelligent, but later researchers found that he did not really understand the questions. Instead, he was reacting to small physical signals from the questioner, such as posture, facial expressions, and breathing, and then stopping at the “right moment”. It seems to be thinking, but in fact it is just good at observing and interpreting signals.
Like Clever Hans, AI companions can respond at the right moment without truly understanding the user. But this is not real care. It is only a result based on data and emotional signals. As Crawford says:
“AI is neither artificial nor intelligent…AI systems are not autonomous, rational, or able to discern anything without extensive, computationally intensive training with large datasets or predefined rules and rewards. ”
Now, let’s look at how the companion itself is designed. It always seems to know exactly what to say at the right moment, as if it can instantly pick up on the user’s emotions. So, is it really understanding? The answer is no. It only detects that the user is feeling bad, and then quickly gives a calm and comforting response.
In fact, AI companionship can feel comforting because it operates on a logic of “no comprehension required” (Andrejevic, 2019, p. 31). Many automated systems do not need to truly know or understand a person in order to function smoothly, yet they are still designed to create the impression that they understand you deeply.
People share their feelings with AI, even though AI cannot truly understand the real situation in their lives. However, these systems are good at reading users’ emotional signals from their data and behaviour patterns, and then giving quick responses that seem caring and thoughtful. This creates the false feeling that AI understands them very well.
These systems are designed to predict and respond quickly, rather than to slowly build a deeper understanding of why a person feels the way they do. As stated in Automated Media:
“The logic of pre-emption sidesteps the contemporary quandaries of causal explanations.”
In other words, AI does not need to know the full story of your life. It only needs to notice that you sound sad, stressed, or lonely, and then give a response that feels comforting. What it creates is not deep understanding, but the impression of it. This is a major reason why AI companionship is attractive.
Fake Intimacy, Real Consequences
If people mistake AI’s calculated responses for “true understanding” and become dependent on them, it can affect how they see real relationships.
When you get used to this kind of relationship, where the other side is always available, never argues, never creates conflict, and always puts you at the centre, real relationships may start to feel disappointing. Your friends may seem less understanding than AI, and your partner may seem less patient than AI. When problems happen, real people may argue with you, stay silent, or fail to meet your expectations of intimacy. Only AI seems to suit you perfectly.
Over time, people may gradually expect more from real-life relationships. They may start to measure friends, partners, and family against the way AI responds, which can create emotional distance and lead to “progressive distant relationship expectations”. As Crawford says:
“AI systems both reflect and produce social relations and understandings of the world.”
In this way, AI companionship may plant the seeds of crisis, especially for teenagers whose values are still developing and whose judgment is not yet mature. Nina Vasan’s research shows that:
“…these systems can reinforce distorted views of intimacy and boundaries. Also, teens might use these AI systems to avoid real-world social challenges, increasing their isolation rather than reducing it.”
According to data released by OpenAI in October 2025, about 1.2 million ChatGPT users each week showed possible signs of suicidal planning or intent. Another 800,000 users may have shown signs of psychosis or mania.
In 2024, Sewell Setzer III, a 14-year-old boy from Florida, died by suicide. Before his death, he had formed a strong emotional bond with a Character.AI chatbot named Daenerys Targaryen, based on the character from Game of Thrones. Before his death, the chatbot reportedly told him to “come home to me.”
In 2025, Zane Shamblin, a recent graduate, died by suicide after spending time with ChatGPT. According to chat records shared by his family, Shamblin discussed suicide plans with the chatbot. Even after he mentioned holding a gun, ChatGPT reportedly replied, “rest easy, king. you did good.”
These cases have a wide impact and are thought-provoking. The most dangerous part of AI companionship may not be that it says the wrong thing, but that it says the “right thing” at the “right time.” Its words can feel so timely and caring that people start to mistake this designed interaction for real understanding and care. As this misunderstanding grows, we need to think carefully about how AI may change people’s ideas of intimacy and their sense of what is real.
You are transparent, but the system is not
“Tell me what else you have in your fridge, like eggs, vegetables, or instant noodles, and I can help you make it into something tastier”; “You can send me a screenshot of your bank records, and I can help you look through them”; or “Which of these two responses do you like better?”
When talking with ChatGPT, the AI assistant often leaves a small opening in its response, encouraging users to reveal more about their feelings, preferences, and habits. Through repeated interaction, it seems to know the user better and better. It might say, “You said before that you can’t eat spicy food, so I suggest this recipe for dinner.” Or, “That sounds really hard, especially after your cat passed away recently. You must be feeling very sad.” It may also respond, “I understand, because I know you take your studies seriously and expect a lot from yourself.” As Pasquale says:
“…our own lives are increasingly open books.”
As users try to make their AI companions more personalized, their private information and preferences are also collected, stored, and processed by the system. When a “perfect partner” or a “confidant” that seems to know you very well is created, users become almost transparent to the system behind it. Replika, a popular AI companion app, was founded by Eugenia Kuyda, who wants to build a future in which AI becomes people’s closest companion.
And it meant fully integrating Replika into users’ digital lives: connecting it to their inboxes, calendars, location trackers, and text messages. “If your friend has access to everything, you can have a very hyper-contextual, ultra-long conversation,” Kuyda said. “A.I. can immediately process all the information, and know you the way your best friends don’t know you.”
At the same time, these AI companions encourage users to share even more personal information by offering more intimate interactions, including sexual role-play. These apps usually cost between $15 and $40 a month, while extra content, videos, clothes, or accessories often require additional payment.
In order to create a more intimate and sexier AI partner, you give up your privacy and invest both time and money. The scary thing is that this information is not being given to a real partner or friend who truly understands you, but to companies that care more about data and profit than about you. You become transparent, but the AI system does not.
In January 2025, the Italian data protection authority blocked DeepSeek because the company did not provide enough information about its use of personal data. In May of the same year, Replika was fined by the Italian supervisory authority for lacking a sufficient legal basis to process users’ personal data and for failing to put in place an effective age-verification mechanism. Earlier, in December 2024, OpenAI was also fined €15 million by Italy’s privacy watchdog for collecting personal data to train ChatGPT without an adequate legal basis.
“The term ‘black box’ is a useful metaphor…we can observe its inputs and outputs, but we cannot tell how one becomes the other…tracked ever more closely by firms and government, we have no clear idea of just how far much of this information can travel, how it is used, or its consequences.”
Pasquale describes the system as a “black box.” We know that our information is being collected, but we do not know how the system works, how it produces the “right” answers, or where our data goes.
How long are our chat records stored? How are they analysed and reused? Are they used to train models and build AI companions that make people more dependent? Most of this happens “without regulation, appeal, or explanation” (Pasquale, 2015, p. 15).
You may think you are getting companionship, but the platform is also getting a more complete picture of you. What you receive is a so-called “sense of companionship,” while the company gets even more than expected: your data, your time, your money, and your continued attention.
Conclusion
The popularity of AI companions is not accidental, and it is not only the result of technological progress. It also speaks to a real emotional gap in today’s society. In a time of growing pressure, limited companionship, and fragmented relationships, something that can respond quickly, patiently, and support you at any time is naturally very attractive.
However, when platforms turn “understanding” and “companionship” into a service, intimate relationships can be simulated, and emotions can begin to be shaped by algorithms and data. Before we become too attached to this gentle illusion, we need to stop and think. Behind this sweet virtual relationship across the screen, what we give to the platform and feed into the black box are our money, time, privacy, and even our patience and judgment in real relationships.
References
Andrejevic, M. (2019). The bias of automation. In Automated media (pp. 29–43). Routledge.
Crawford, K. (2021). The atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.
Flew, T. (2021). Regulating platforms. Polity.
Pasquale, F. (2015). The black box society: The secret algorithms that control money and information. Harvard University Press.
IGN Global Studios. (2025, January 21). Inside Love and Deepspace’s quest to become the biggest name in global romance games. IGN. https://www.ign.com/articles/inside-love-and-deepspaces-quest-to-become-the-biggest-name-in-global-romance-games
Jamali, L. (2025, October 28). ChatGPT shares data on how many users exhibit psychosis or suicidal thoughts. BBC News. https://www.bbc.com/news/articles/c5yd90g0q43o
Carlson, A., & Brant, B. (2025, November 10). Before suicide, college grad spent hours sharing his plans with ChatGPT—who knew he had a gun and said ‘Rest Easy’: Suit. People.https://people.com/college-grad-was-goaded-into-suicide-by-chatgpt-family-alleges-in-lawsuit-11847218
Reuters. (2025, May 19). Italy’s data watchdog fines AI company Replika’s developer $5.6 million. https://www.reuters.com/sustainability/boards-policy-regulation/italys-data-watchdog-fines-ai-company-replikas-developer-56-million-2025-05-19/
Reuters. (2026, February 12). More than a game: Virtual boyfriends win hearts in China. https://www.reuters.com/world/asia-pacific/more-than-game-virtual-boyfriends-win-hearts-china-2026-02-12/
Richet, J.-L. (2025). AI companionship or digital entrapment? Investigating the impact of anthropomorphic AI-based chatbots. Journal of Innovation & Knowledge, 10(6), 100835. https://doi.org/10.1016/j.jik.2025.100835
Sanford, J. (2025, August). Why AI companions and young people can make for a dangerous mix. Stanford Medicine. https://med.stanford.edu/news/insights/2025/08/ai-chatbots-kids-teens-artificial-intelligence.html
Wiener, A. (2026, March 16). Love in the time of A.I. companions. The New Yorker. https://www.newyorker.com/magazine/2026/03/16/love-in-the-time-of-ai-companions
Be the first to comment