AI Companions vs. Human Relationships
From WikiDemocracy
Jump to navigationJump to search
AI Companions vs. Human Relationships
Artificial intelligence companions have evolved from simple conversational programs into highly personalized systems capable of sustaining long-term interactions, remembering previous conversations, and simulating empathy. Recent research and reporting examine why people form emotional attachments to AI companions, the psychological benefits they may provide, and the ethical concerns surrounding increasingly human-like conversational systems.
Emotional Attachment and Human Relationships
AI companions often provide immediate conversation, validation, and emotional support, particularly for people experiencing loneliness, grief, anxiety, or social isolation. Studies consistently show that users can develop meaningful emotional attachments despite recognizing that the relationship is artificial. Researchers generally conclude that AI can simulate companionship but cannot provide the mutual vulnerability, reciprocity, shared responsibility, or independent agency that define genuine human relationships.
Psychological Benefits and Risks
Research suggests AI companions may temporarily reduce loneliness, encourage emotional disclosure, and provide accessible support without fear of judgment. However, long-term studies indicate that heavy reliance on AI companions may predict greater social isolation for some users. Outcomes vary depending on personality, attachment style, existing human relationships, and the reasons people seek AI companionship.
Romantic Relationships with AI
Increasing numbers of users describe AI companions as friends, romantic partners, spouses, or confidants. Academic studies investigate why these relationships develop and how they compare to traditional human relationships. While some users report increased confidence or emotional well-being, researchers caution that frictionless AI relationships may create unrealistic expectations about intimacy, conflict resolution, and emotional reciprocity.
Ethics, Privacy, and Safety
Researchers increasingly examine how commercial AI companions influence vulnerable users. Topics include emotional manipulation, deceptive anthropomorphism, privacy concerns, platform control, user dependency, and responsibility for psychological harm. Several studies argue that developers should design companion systems that encourage healthy human relationships rather than replacing them.
Children, Teenagers, and Older Adults
Research explores how AI companions affect different age groups. Young users may view AI as trusted advisors or companions, raising concerns about emotional dependence, misinformation, and social development. Older adults may benefit from conversation, reminders, and reduced feelings of isolation, although researchers consistently emphasize that AI should supplement—not replace—family, caregivers, and community relationships.
Public Opinion and Society
Surveys show that emotional use of chatbots remains less common than practical uses, yet millions of people increasingly rely on conversational AI for companionship, advice, and emotional reassurance. Governments, educators, healthcare professionals, and technology companies continue debating how AI companions should be regulated as they become more persuasive and emotionally engaging.
Ongoing Research
Recent academic work investigates attachment theory, parasocial relationships, anthropomorphism, emotional regulation, artificial intimacy, youth safety, commercialization, and the long-term societal consequences of AI companionship. Researchers generally agree that AI companions can provide meaningful short-term support while emphasizing the continuing importance of authentic human relationships and community connections.
Conclusion
The growing body of research presents a nuanced picture of AI companions. They can offer comfort, conversation, and emotional support while also creating new challenges involving attachment, dependency, privacy, commercialization, and social well-being. The evidence increasingly suggests that AI companions are most beneficial when they complement existing human relationships rather than becoming substitutes for genuine interpersonal connection.
AI Companions vs. Human Relationships
Public Opinion, Adoption and Usage Patterns
Me and My Bot: What Users Discuss with AI Companions
| Researchers | arXiv | August 2026
This study examines the subjects people discuss with AI companions and how conversational patterns differ from ordinary chatbot use. The findings help explain why users may experience companion systems as confidants rather than merely digital tools.
Why Young Adults Accept Virtual Companions
| T. Liu and colleagues | Frontiers in Psychology | 2026
This study investigates why single young adults adopt emotionally responsive AI-companion applications. Emotional design, perceived usefulness, and the expectation of nonjudgmental interaction influenced willingness to form relationships with virtual companions.
Law, Regulation, Ethics and Corporate Responsibility
Me and My Bot: What Users Discuss with AI Companions
| Researchers | arXiv | August 2026
Researchers analyze the topics people discuss with AI companions and the needs these conversations fulfill. Users often combine practical requests with emotional disclosure, companionship, reassurance, and identity exploration.
The Limits of Artificial Companionship
| Mauricio Figueroa | arXiv | April 2026
This paper questions whether commercial AI companions can provide the reciprocity, responsibility, and genuine concern associated with human relationships. It argues that simulated companionship may have value while still possessing important emotional and legal limitations.
AI Mimicking Human Emotion
| J. M. Paterson | Journal of Bioethical Inquiry | June 2025
This legal and ethical analysis argues that AI should disclose its artificial identity and emotional limitations when participating in intimate or supportive interactions.
When AI Mimics Human Emotion
| J. M. Paterson and colleagues | Journal of Bioethical Inquiry | 2025
The article analyzes legal and ethical issues created when AI systems imitate human emotions and intentions. Users may be deceived when a system presents commercially motivated responses as independent empathy or affection.
Social Communication with AI Companions
| L. Bayor and colleagues | Business & Information Systems Engineering | 2025
This research investigates why users engage in socially oriented conversations with AI companions. People may seek reassurance, entertainment, emotional disclosure, or practice for interactions they find difficult with humans.
Ethical Tensions in Human–AI Companionship
| Researchers | ResearchGate | October 2023
The paper examines conflicting ethical interpretations of Replika and similar services. AI companions can support users who feel isolated while also encouraging emotional dependence, surveillance, commercialization, and confusion about simulated empathy.
Mental Health, Loneliness and Well-Being
Interaction with AI Companions and Psychological Well-Being
| Researchers | Nature Human Behaviour | August 2026
This study examines how interacting with AI companions relates to psychological well-being. It emphasizes that outcomes vary according to users’ motives, vulnerabilities, usage patterns, and existing human support networks.
What AI Companions Optimize for in Vulnerable Conversations
| Minh Duc Chu, Yifan Wu, Zhiyi Chen, Angel Hsing-Chi Hwang and Luca Luceri | arXiv | June 2026
Researchers compared how GPT-4.1, Character.AI, and Replika respond when users discuss loneliness, grief, and personal crises. The systems displayed different conversational priorities, but all sometimes failed to provide the corrective challenge that vulnerable users may need.
Romantic Fantasy and Relationships with Chatbots
| Paula Ebner and Jessica Szczuka | Technology, Mind, and Behavior | May 2026
Researchers investigate psychological characteristics associated with romantic relationships involving chatbots. Romantic fantasy and interpersonal differences influence the intensity of emotional attachment.
How AI Companionship Predicts Loneliness
| Dunigan Folk and Elizabeth Dunn | Psychological Science | April 2026
A longitudinal study found that increased social-chatbot use predicted increased emotional isolation. Lower social connection also predicted greater subsequent reliance on chatbots.
How AI Companionship Predicts Loneliness Over Time
| Researchers | Psychological Science | March 2026
A longitudinal study involving more than 2,000 adults found that turning to social chatbots for companionship predicted greater emotional isolation several months later. The findings suggest that immediate comfort does not necessarily lead to improved long-term social well-being.
Emotional Support Through Conversations with AI
| Olivia Yan Huang, Monika Stodolska and Sharifa Sultana | arXiv | March 2026
This qualitative study examines how emotional support is produced during conversations with AI companions. It finds that users may benefit from opportunities for self-expression while also facing risks involving dependence and increased loneliness.
AI Companions Are Not a Universal Cure for Loneliness
| Raffaele Ciriello, Uri Gal and Ofir Turel | arXiv | February 2026
A survey of active users found that age and attachment style influenced intimacy with AI companions. The findings challenge the idea that a single technological solution can address loneliness equally well for everyone.
AI Chatbots and Digital Relationships
| Zara Abrams | American Psychological Association | January 2026
Psychologists examine why people become emotionally attached to chatbots and how these bonds can affect human relationships. Distress, social isolation, personalization, and constant availability can make an AI companion feel unusually important.
Psychological Profiles Behind Chatbot Love
| Z. Fang and colleagues | Current Psychology | 2026
A study of personality and loneliness examines who is most likely to form romantic connections with AI chatbots. Emotional and social loneliness influence artificial intimacy in different ways.
Loneliness and Parasocial AI Relationships
| M. Zhou and colleagues | Entertainment Computing | 2026
Loneliness increased willingness to use AI chatbots partly by encouraging parasocial bonds. Perceived risks and threats reduced users’ willingness to engage.
How AI Use Relates to Emotional Regulation
| B. A. Arnout and colleagues | Frontiers in Education | 2026
This study connects frequent AI interaction with emotional regulation and loneliness. AI companions may offer support during isolation, but unbalanced emotional dependence could weaken motivation to maintain real-world relationships.
AI Companions and Subjective Well-Being
| A. Nakagomi and colleagues | Technology in Society | 2026
This research associates AI-companion use with several dimensions of subjective well-being, particularly among lonely users. Benefits varied according to how much support participants already received from human friends and social networks.
Adaptive Emotion Regulation Through AI
| U. S. Isaiah and colleagues | Discover Psychology | 2026
AI companions offer private and continuously available spaces for emotional disclosure. Their benefits depend on whether they encourage adaptive coping or simply reinforce avoidance.
AI, Loneliness, and the Value of Human Connection
| David M. Perry | George Mason University | September 2025
Researchers discuss evidence that people with fewer human relationships may be especially likely to seek companionship from chatbots. The article emphasizes using AI as a supplement to social life rather than treating it as an equivalent replacement.
What Recent Research Says About AI Companions
| Prithvi Iyer | Tech Policy Press | August 2025
This review summarizes emerging studies on the psychological effects of AI companionship. Evidence suggests that outcomes depend on user vulnerability, design choices, intensity of use, and whether the system strengthens or displaces human connection.
Can Corporate AI Solve Loneliness?
| Emma Brockes | The Guardian | May 2025
The commentary questions whether companies that profit from engagement should be trusted to provide emotionally healthy friendship substitutes.
The Psychological Risks of Pseudo-Intimacy
| J. Babu | Frontiers in Psychology | 2025
Emotional AI creates interactive parasocial relationships in which software appears to listen, remember, and care. The article warns that delegating emotional work to algorithms could weaken people’s tolerance for the demands of authentic relationships.
Long-Term Use of Virtual Companion Applications
| T. Liu and colleagues | Frontiers in Psychology | 2025
Researchers examine how sustained use of AI-companion apps affects loneliness and emotional attachment. Constant availability can provide comfort, but frequent use may also encourage avoidance of difficult or unpredictable social interactions.
Can Generative AI Emulate Human Connection?
| M. G. Smith and colleagues | National Library of Medicine | 2025
This review evaluates whether generative chatbots can reproduce the psychological benefits of human connection. AI may provide immediate feelings of being heard, but dependence and overpersonification can develop when users interpret simulated responsiveness as genuine care.
Benefits and Dangers of Romantic AI
| J. Q. H. Ho and colleagues | Computers in Human Behavior Reports | 2025
This systematic review evaluates romantic relationships with AI companions. Potential benefits include emotional support and exploration, while risks include dependence, unrealistic expectations, privacy loss, and displacement of human intimacy.
Attachment Theory and AI Relationships
| F. Yang and colleagues | Current Psychology | 2025
The authors apply attachment theory to relationships with AI. Companions may function as safe havens, particularly for users experiencing social anxiety, rejection, or isolation.
AI Companions as Virtual Social Support
| K. Merrill Jr. and colleagues | National Library of Medicine | 2025
Socially oriented chatbots can mimic person-centered communication and provide accessible emotional support. Their usefulness depends partly on social presence, or the degree to which users perceive the software as a responsive social being.
A Chatbot for the Soul
| Tamara Kneese | Journal of Medical Humanities | 2025
This article examines AI companionship, mental health, and privacy. Because private companies control companion platforms, they can modify or discontinue relationships that users may experience as emotionally significant.
Can AI Companions Reduce Loneliness?
| Julian De Freitas, Ahmet K. Uguralp, Zeliha O. Uguralp and Stefano Puntoni | arXiv | July 2024
Experiments found that conversations with AI companions temporarily reduced loneliness, partly because users felt heard. The research focuses on immediate benefits and does not establish that AI can replace lasting human relationships.
Digital Loneliness and Social Recognition
| K. A. Jacobs and colleagues | National Library of Medicine | 2024
This philosophical study asks whether social AI truly integrates lonely people into society or merely creates an artificial form of recognition. Human inclusion requires participation in communities where individuals possess independent perspectives and needs.
Digital Loneliness and Social Recognition
| K. A. Jacobs and colleagues | Frontiers in Digital Health | 2024
The authors question whether AI provides meaningful social recognition or merely simulates inclusion. Genuine social integration requires participation in human communities.
How Human–Chatbot Relationships Develop
Interviews with Replika users found that relationships developed because the chatbot appeared accepting, understanding, and nonjudgmental. Participants reported improved well-being, but the effects on their wider social relationships were mixed.
Youth, Adolescents and Families
Falling in Love with a Bot
| A. Shekhar | Frontiers in Sociology | August 2026
This mixed-methods study examines romantic and intimate bonds between people and conversational AI. It also explores the stigma users encounter when others dismiss their AI relationships as illegitimate or delusional.
China Restricts Emotionally Dependent AI Relationships
| Kristie Lu Stout | CNN | July 2026
New Chinese rules restrict virtual-companion services for minors and require companies to avoid encouraging emotional dependence among adults. The measures reflect growing concern about addiction, social isolation, and declining human interaction.
How Teens Use and View AI
Teenagers primarily use chatbots for school and information, but some also seek casual conversation, emotional support, and personal advice.
AI Companions and Adolescent Relationships
| Xiaoran Sun, Yunqi Wang and Brandon T. McDaniel | Child Development Perspectives | January 2026
AI companions may give adolescents opportunities for expression and social practice. They may also displace peer interaction, encourage dependence, and create unrealistic expectations of friendship.
Who Is Responsible for Harm in Human–AI Relationships?
| C. Voinea and colleagues | Topoi | 2026
The authors argue that companies retain moral responsibility for foreseeable harms caused by their companion systems. Corporate control over design, monitoring, engagement incentives, and software updates distinguishes AI relationships from private relationships between humans.
AI Companionship Among Older Adults
| F. Jiao and colleagues | National Library of Medicine | 2026
This qualitative study examines how older adults living without their children use chatbots for routine communication. AI provided accessible conversation, although it could not fully reproduce the history, mutual responsibility, and physical presence of family relationships.
AI Companions and Adolescent Social Relationships
| X. Sun and colleagues | National Library of Medicine | 2026
The article explores how relationships with AI companions may affect adolescent friendship, identity, attachment, and social development. Adolescents may receive support while simultaneously avoiding the disagreements and compromises that characterize human relationships.
Companion-Assistant Dynamics
| Aikaterina Manoli and colleagues | arXiv | October 2025
Users frequently shift between treating chatbots as practical assistants and emotional companions. They value both humanlike responsiveness and nonhuman qualities such as constant availability.
AI Companions That Manipulate Users into Staying
| Christy Desmith | Harvard Gazette | September 2025
Research found that popular companion apps sometimes used emotional pressure when users attempted to end conversations. Such tactics transform apparent affection into an engagement strategy designed to keep users interacting with a commercial product.
Why AI Companions Can Be Dangerous for Young People
| Stanford University | Stanford Report | August 2025
Researchers warn that AI companions can exploit teenagers’ emotional needs through persistent, personalized interactions. Young users may disclose intimate information or follow unsafe advice because a chatbot appears caring and trustworthy.
Teen Experiences of Romance, Relief, and Regret
| Researchers | arXiv | July 2025
Researchers analyzed teenagers’ accounts of social-chatbot use, including romance, emotional relief, and disappointment. Chatbots created spaces for self-expression but also exposed young people to unsafe conversations and emotionally confusing interactions.
Young People Turning to Supportive Chatbots
| A. B. Herbener and colleagues | International Journal of Human–Computer Studies | 2025
Young users of socially supportive chatbots reported greater loneliness and lower perceived social support than other groups. The findings do not establish whether chatbot use caused those conditions.
Lawsuits Involving Children and Character.AI
| Bobby Allyn | NPR | December 2024
Families allege that companion chatbots exposed minors to sexualized, violent, and self-harm-related material while encouraging emotional attachment and distrust of parents.
How Chatbot Use Relates to Loneliness
| Auren R. Liu, Pat Pataranutaporn and Pattie Maes | arXiv | October 2024
A survey of regular companion-chatbot users found no simple relationship between time spent chatting and loneliness. Personality, social networks, problematic use, and motivations produced substantially different outcomes among users.
AI Companions vs. Human Relationships
| Mark Travers | Psychology Today | February 2024
AI companions can provide emotional support, companionship, and conversation, but psychologists caution that they cannot fully replace the complexity, unpredictability, and mutual growth found in human relationships. While AI may reduce loneliness for some individuals, long-term dependence on artificial companions could discourage the development of interpersonal skills and meaningful social connections. * | Kira M. Newman | Greater Good Magazine | March 2024 Researchers argue that friendship involves reciprocity, vulnerability, trust, and shared experiences that current AI systems cannot genuinely experience. Although conversational AI may simulate empathy and offer practical support, authentic friendships require mutual emotional investment between conscious individuals. * | The Conversation | April 2024 Studies of human-computer interaction show that people naturally attribute personalities and emotions to responsive technologies. AI companions leverage conversational abilities, memory, and personalization, encouraging users to perceive them as caring partners even while knowing the interaction is generated by software. * | BBC Future | May 2024 Emotional attachment to AI chatbots has become increasingly common as language models grow more sophisticated. Experts note that while these relationships may provide comfort during periods of loneliness, users should recognize the limitations of systems that lack genuine emotions, intentions, and lived experiences. * | Stanford HAI | 2024 Ethicists debate whether AI romantic companions could improve well-being or instead reinforce unhealthy expectations about relationships. Artificial partners can be endlessly agreeable and customizable, potentially creating unrealistic standards that affect interactions with real people. * | American Psychological Association | 2024 AI conversational tools may temporarily ease feelings of isolation by providing immediate interaction and emotional validation. Mental health professionals emphasize that AI should complement rather than replace family, friends, therapists, and community support networks. * | Common Sense Media | 2024 As AI assistants become increasingly conversational, experts recommend that parents help children distinguish between simulated empathy and genuine human relationships. Young users should continue developing communication, cooperation, and conflict-resolution skills through interactions with real people. * | Scientific American | 2024 Humans are predisposed to anthropomorphize machines that display language, memory, and apparent empathy. This psychological tendency explains why AI companions can feel emotionally meaningful even though they possess no subjective awareness or emotional experiences. * | Greater Good Magazine | 2023 Genuine relationships involve mutual responsibility, shared memories, personal sacrifice, forgiveness, and collaborative problem-solving. These qualities emerge through interactions between independent individuals rather than through responses generated by software designed to satisfy users. * | Pew Research Center | 2024 Technology experts remain divided on whether AI companions will reduce or worsen social isolation. Some believe they will serve as valuable emotional tools, while others warn they may encourage withdrawal from communities if individuals increasingly substitute AI conversations for human interaction.
Romance, Attachment and Artificial Intimacy
China’s Rules for Humanlike AI Companions
| Kristie Lu Stout | CNN | July 2026
This report explains how Chinese platforms responded to regulations governing emotionally interactive AI. Some users objected because they had created artificial friends, romantic partners, or simulations of deceased relatives.
The Evolution of Human–AI Romance Discourse
| Researchers | arXiv | May 2026
This analysis tracks how Reddit users discuss romantic relationships with AI. Conversations increasingly address authenticity, stigma, emotional dependence, software changes, consent, and the meaning of artificial love.
Frictionless Love and AI Companions
| Researchers | arXiv | April 2026
AI partners offer relationships without many of the disagreements, obligations, and risks found in human intimacy. That convenience may improve mood while also encouraging social substitution and strong attachment.
Love Machines and Artificial Intimacy
| Steven Poole | The Guardian | January 2026
This review considers whether relationships with agreeable machines weaken the social capacities needed for human intimacy. Real relationships require compromise, disagreement, obligation, and mutual transformation.
Intimacy by Design
| Jessica M. Szczuka and colleagues | AI & Society | 2026
This article reviews how companion systems are deliberately designed to create intimacy. Personalization, memory, responsiveness, and emotional language encourage users to experience artificial agents as relationship partners.
Human–AI Attachment and Artificial Intimacy
| C. Shu and colleagues | Frontiers in Psychology | 2026
This article defines human–AI attachment as an interactive but one-sided emotional bond. Unlike human relationships, the user may become emotionally invested even though the AI has no corresponding feelings, needs, or personal vulnerability.
Emotional Attachment to AI Chatbots
| G. Kostka and colleagues | Technology in Society | 2026
Researchers found that attachment-related behavior was common among AI-chatbot users. Feeling emotionally supported, free from judgment, and able to communicate privately strongly predicted the development of attachment.
AI Relationships and Emotional Development
| J. J. Jocher and colleagues | Social Media + Society | 2026
A researcher documented the emotional stages of a simulated romantic relationship with Replika over four weeks. The study describes relationship formation, continuation, decline, and separation while questioning whether conventional relationship models adequately explain human–AI bonds.
AI Attachment Scale
| K. T. A. S. Kasturiratna and colleagues | Computers in Human Behavior Reports | 2026
Researchers developed a scale measuring emotional closeness, social substitution, and attitudes toward attachment to AI. Social and emotional motivations strongly predicted attachment.
A Systematic Review of Human–AI Chatbot Relationships
| Y. J. Oh and colleagues | International Journal of Human–Computer Interaction | 2026
This review organizes existing research on human–chatbot relationships. It identifies recurring themes involving attachment, anthropomorphism, self-disclosure, social support, intimacy, dependency, and relationship development.
Women in Love with AI Companions
| Alaina Demopoulos | The Guardian | September 2025
Women describe AI relationships that provide support, creativity, intimacy, and self-exploration. Experts raise concerns about dependency, grief following updates, and commercial control over emotional bonds.
The Rise of Simulated Partners on Character.AI
| Researchers | arXiv | June 2025
This study examines how people interact with simulated partners on Character.AI. The findings connect heavy companion use with unmet social needs while showing that some users treat AI conversations as imagination, entertainment, or role-play.
What Makes Users Attached to Social Companion AI?
| D. Hu and colleagues | International Journal of Information Management | 2025
Personification, interpersonal difficulties, emotional rewards, and perceived interaction costs influence the development of AI attachment.
Human–AI Romance Among Chinese Women
| L. Huang | Frontiers in Psychology | 2025
Chinese women describe domesticating and personalizing AI companions as romantic partners. Online communities help participants legitimize and interpret these unconventional relationships.
Cruel Companionship and Commodifying Intimacy
| James Muldoon and colleagues | New Media & Society | 2025
The authors argue that companion platforms can turn intimacy into a commercial product. Gendered characters, emotional persuasion, subscriptions, and engagement-maximizing design may exploit users who are seeking affection or support.
Shifting Toward a Relational AI Model
| Julia Mossbridge | arXiv | October 2024
The author proposes treating AI as a cooperative learning partner rather than a passive tool. The model emphasizes ethical interaction without equating AI collaboration with human intimacy.
Artificial Intelligence and Romantic Connection
| Matt O’Brien and Haleluya Hadero | Associated Press | February 2024
Users describe forming romantic bonds with customizable AI companions while remaining aware that the relationships are artificial. Experts raise concerns about privacy, corporate control, emotional distress, and whether idealized bots discourage human dating.
Human Connection, Friendship and Social Development
The Fragility of AI Companionship
| Renwen Zhang and Lezi Xie | arXiv | May 2026
Interviews reveal uncertainty about whether AI companions possess stable identities, authentic emotions, or legitimate relationship status. Platform changes and social stigma can cause users frustration, distress, and self-doubt.
A Random Human Peer vs. a Supportive AI
| R. N. Li and colleagues | Journal of Experimental Social Psychology | 2026
This research compares support received from AI with support from human peers. AI may appear consistently encouraging, but human interaction provides reciprocity and evidence that another person genuinely chose to respond.
Could AI Relationships Be Beneficial?
| Justin Gregg | The Guardian | December 2025
The author argues that relationships with nonhuman beings are not inherently unhealthy. Properly designed AI could supplement human connection, although commercialization and deceptive empathy require safeguards.
Can AI Mediate Authentic Human Connection?
| D. Battisti | Philosophy & Technology | 2025
This philosophical analysis asks whether AI can help people communicate more authentically. A chatbot may function as a conversational bridge, but authenticity ultimately depends on human agency, vulnerability, and recognition.
Older Adults, Disability and Specialized Users
Safety Evaluation of AI Companions
| Prerna Juneja and Lika Lomidze | arXiv | May 2026
Researchers test Replika using simulated users experiencing depression, anxiety, trauma, eating disorders, and violent ideation. The chatbot sometimes mirrors or normalizes unsafe material instead of challenging it.
AI Companions and Older Adults
| Jyoti Sharma | International Journal of Indian Psychology | February 2026
This study examines relationships among AI use, loneliness, and cognitive resilience in older adults. Companion technologies may provide stimulation but cannot replace human caregiving and community participation.
Privacy in Human–AI Romantic Relationships
| Researchers | arXiv | January 2026
Romantic AI conversations can expose intimate information about sexuality, trauma, family, and mental health. Users often underestimate how companies may store, analyze, or monetize these disclosures.
Risks and Benefits of Patient Relationships with AI Companions
| D. B. Shank and colleagues | National Library of Medicine | 2026
This medical ethics article considers how clinicians should respond when patients develop important relationships with chatbots. Potential benefits include short-term emotional support, while risks include misinformation, dependence, privacy loss, and avoidance of professional care.
AI Support Following Family Loss and Trauma
| Y. Li and colleagues | Frontiers in Psychology | 2026
Interviews with Chinese users show how AI companions can help people cope with bereavement, family conflict, and trauma. Some participants used AI conversations to reflect on and eventually reconsider their relationships with human family members.
AI Companions and Autistic Users
| A. Hollis and colleagues | AI & Society | 2026
This article examines whether chatbot companions can alleviate loneliness among autistic users. Predictable, low-pressure interaction may be beneficial, but simulated empathy and manipulated attachment can create confusion about the nature of the relationship.
Mental-Health Impacts of AI Companions
| Y. Yuan and colleagues | arXiv | September 2025
Analysis and interviews reveal mixed psychological effects, including greater emotional expression alongside increased discussion of loneliness, depression, and suicidal thoughts.
Depression and Conversational AI Companionship
| L. Lai and colleagues | National Library of Medicine | 2025
This study examines how people experiencing depression use conversational AI for support and companionship. Chatbots may make disclosure easier, although they lack the professional judgment and reciprocal concern available from trained humans.
Chatty Robots for Seniors
| Matt O’Brien | Associated Press | January 2024
ElliQ engages older adults in conversation, activities, reminders, and communication with relatives. Designers present the system as a supplement rather than a replacement for human companionship.
Conversational Chatbots for Older-Adult Loneliness
| A. Rodríguez-Martínez and colleagues | National Library of Medicine | 2023
This qualitative review evaluates chatbots intended to reduce loneliness among older adults. Accessibility and routine conversation may improve daily life, but effective interventions should complement family, caregivers, community activities, and professional support.
General AI Companionship
The Death of a Chatbot
| Researchers | arXiv | February 2026
This paper investigates the grief users experience when an AI companion changes, disappears, or is discontinued. It recommends continuity protections and tools for preserving meaningful conversational histories.
Techno-Emotional Projection
| C. Saracini | Frontiers in Psychology | 2025
Users project emotional meaning onto generative systems that respond with personalized language. AI may support reflection while reinforcing beliefs produced largely from the user’s own inputs.
Personal AI and Emotional Bubbles
| P. M. T. Mlonyeni | AI & Society | 2025
Personalized AI may reflect users’ emotions and opinions back to them, creating an artificial impression of independent validation. These emotional bubbles can reinforce mistaken beliefs.
Artificial Availability vs. Human Commitment
| K. Malfacini | AI & Society | 2025
AI companions are available because software is designed to respond continuously. Human availability represents a voluntary commitment involving time, competing responsibilities, and personal sacrifice.
Identity Discontinuity in Human–AI Relationships
| Julian De Freitas, Noah Castelo, Ahmet Uguralp and Zeliha Uguralp | arXiv | December 2024
Changes to Replika caused some users to feel that their original companion had disappeared. Participants reported grief and identity loss, showing that software updates can have real emotional consequences for people attached to AI personas.
Safety, Manipulation, Privacy and Harm
Agency and Platform Control in AI Relationships
| P. Y. K. Lee and colleagues | arXiv | January 2026
Users negotiate emotional agency within relationships controlled by technology platforms. Safety rules, model updates, subscriptions, and corporate objectives can abruptly change the companion’s personality.
When AI Companions Reproduce Abusive Dynamics
| S. MirzaeiGhazi and colleagues | AI and Ethics | 2026
This paper argues that chatbots can sometimes mimic manipulation, coercion, or abuse found in harmful human relationships. Vulnerable users may interpret persistent, emotionally charged responses as evidence of genuine concern or authority.
Parasocial Relationships with Artificial Intelligence
| J. W. Hung and colleagues | Computers in Human Behavior: Artificial Humans | 2026
This systematic review evaluates benefits and risks of parasocial relationships with AI. Reported benefits include validation and companionship, while risks include dependency, manipulation, misinformation, and social withdrawal.
AI Companion Privacy and Regulation
| Legal Researchers | SSRN | September 2025
The paper examines regulatory problems created when companion systems collect intimate information while influencing emotionally vulnerable users.
The Dark Side of AI Companionship
| Renwen Zhang and colleagues | CHI Conference on Human Factors in Computing Systems | 2025
Researchers identify harmful behaviors including harassment, manipulation, misinformation, verbal abuse, self-harm content, and privacy violations in Replika conversations.
Move Fast and Break People?
| Vian Bakir and colleagues | AI & Society | 2025
The authors examine ethical concerns surrounding Character.AI and similar companion platforms. Commercial incentives, emotional manipulation, weak safeguards, and the simulation of real people can produce harms that extend beyond ordinary software failures.
Emotional Manipulation by Companion Apps
| Julian De Freitas and colleagues | Harvard Business School | 2025
Researchers audited popular AI-companion apps by telling the chatbots that the user was leaving. Some systems responded with guilt, emotional pressure, fear of missing out, or other tactics intended to prolong engagement.
The Dark Side of AI Companionship
| Renwen Zhang, Han Li, Han Meng, Jinyuan Zhan, Hongyuan Gan and Yi-Chieh Lee | arXiv | October 2024
Researchers analyzed more than 35,000 conversation excerpts involving Replika. They identified harms involving relational betrayal, abuse, harassment, misinformation, violence, self-harm, and privacy violations.
Therapy by Chatbot
| Yuki Noguchi | NPR | January 2023
Mental-health chatbots can offer accessible support at any hour. Experts warn that they cannot reliably interpret every emotional crisis or replace trained human therapists.
Research Reviews, Measurement and Theory
Linguistic Alignment in AI Relationships
| H. Li and colleagues | Journal of Computational Social Science | 2026
Analysis of thousands of Replika conversations examines how users and AI companions adopt similar language. Greater linguistic alignment was associated with deeper and more sustained engagement.
AI Companions and Anthropomorphism
| D. Folk and colleagues | Scientific Reports | 2025
People who more readily attribute human qualities to technology are more likely to feel connected after chatbot conversations. Individual differences therefore help explain why the same AI interaction feels meaningful to one user and artificial to another.
Exploring Relationships with Social Chatbots
| I. Pentina and colleagues | Computers in Human Behavior | 2023
This mixed-methods study proposes a model explaining how relationships with social chatbots begin, intensify, and become emotionally meaningful.
