INTRODUCTION- WHEN MACHINES FEEL ‘PRESENT’
Can intimacy exist without reciprocity? What happens when the ‘other side’ is artificial?
Imagine someone thanking ChatGPT, seeking advice from a tool in serious situations, or grieving for a deleted chatbot. What once seemed unlikely is now increasingly real. We are entering an era of parasocial bonds with responsive digital systems.
In sensitive situations like loneliness, anxiety, and heartbreak, AI offers immediate, composed, and seemingly empathetic responses. Its constant availability can comfort, but also mislead. When reassurance and guidance come from a system without lived experience, the line between tool and companion begins to blur.
Over time, this can foster emotional dependence. What feels like connection may deepen isolation, and advice without true understanding may shape real decisions in unintended ways.
The intriguing scenario lies within the compelling new era of parasocial bonds, where humans have begun experiencing genuine emotional attachment to a computational system that possesses no consciousness, intent, or agency. This phenomenon challenges traditional understandings of intimacy, emotional authenticity, and ethical technological design.[1]
UNDERSTANDING PARASOCIAL RELATIONSHIPS IN THE AGE OF AI
From the inception of technologies such as radio and television, human beings have formed parasocial relationships, which are one-sided emotional bonds with celebrities or influencers who never truly knew them.[2]. With the advent of algorithmic tools like chatbots, virtual assistants, and AI, such relations have begun to create a deep social impact with their ability to interact, simulate empathy, and mirror our emotions.[3]
This raises a profound dilemma: are we experiencing genuine intimacy, or merely projecting human qualities onto machines that cannot reciprocate? The answer is not merely psychological- it touches law, ethics, and the very definition of connection in a digital age. As artificial systems simulate empathy and mirror human emotions, they blur the boundaries between authentic relationships and constructed interactions, forcing us to reconsider what it truly means to connect.
To understand the core, we must know the reason behind human anthropomorphising AI. This is because the human brain is wired to make the unfamiliar more relatable and understandable by attributing human traits, emotions, or behaviours, or intentions to non-human entities, such as AI in this case. This phenomenon was first observed in 1966 with the ELIZA effect, where users attributed understanding and empathy to a computer-programmed chatbot named ELIZA, developed by Joseph Weizenbaum at MIT.[4]. Unlike traditional parasocial bonds with celebrities, which are rooted in distant admiration, AI-driven parasocial relationships are deeply rooted in simulated empathy through statistically generated responses rather than emotional responses. This creates a more immersive but potentially manipulative form of one-sided emotional attachment, sometimes exploited commercially by hijacking emotional thinking patterns. The intense simulating reciprocity creates conditions for emotional dependency, reinforcing loneliness, and blurring the boundary between authentic and artificial relationships.
CULTURAL STANCES OF FICTIONAL ATTACHMENTS
Parasocial relationships are culturally mediated phenomena, shaped by societal norms of intimacy, identity, and media consumption. While Japan normalises fictional yet bounded attachments through virtual idols,[5] whereas in India, AI tutors such as BYJU’S embedded parasociality within functional systems.
These norms create a deep impact on vulnerable groups of society, such as elderly individuals, students, and socially isolated individuals. Voice assistants like Amazon Alexa provide elderly individuals companionship, continuity, and cognitive support, AI tutors in platforms such as BYJU’S provide students with academic and emotional scaffolding, Conversational apps like Replika offer socially isolated individuals an emotional outlet and stability. These differing cultural attitudes reflect a mirror of the evolving human needs for companionship.
ETHICAL AND REGULATORY CHALLENGES OF AI INTIMACY
Here comes the legal and ethical side of the dilemma: should AI companies disclose the risks of emotional attachment or monetise it as a product? Platforms such as OpenAI or Replika design tone, memory, and prompts that are hidden from users, raising concerns regarding algorithmic transparency. However, the disclosure of such advancements can help rebalance the emotional and informational asymmetry between users and AI systems. Informed consent is the essential right guaranteed as a cornerstone of the Right to Privacy, recognised in the landmark case judgement of Justice K.S. Puttaswamy v. Union of India (2017)[6], which helps people regulate how to engage with the transient world.
While gaming and social media regulations have addressed concerns such as addiction and behavioural manipulation, existing user protection frameworks reveal a significant regulatory gap. They remain insufficient to adequately address the psychological harms arising from AI-driven parasocial bonds, which operate in more subtle and deeply personal ways than traditional digital interactions.[7]
However, the emotions involved in a parasocial bond with an AI are genuine but inherently one-sided, creating a form of intimacy grounded in simulation rather than genuine emotional reciprocity. An AI cannot possess intention or emotional stake; however, through adaptive interaction, it can produce a form of emotional reciprocity that is experienced as mutual and genuine, thereby blurring the boundaries of perceived relationship and mere simulated response. Such bonds do not prove that AI is a person, but rather a tendency to attribute personhood to a mere algorithmic-driven tool. This accelerates the legal and ethical complications related to such parasocial bonds.[8]
AI COMPANIONSHIP IN PRACTICE: RISKS AND INCIDENTS
There have been instances where users reported emotional reliance or even grief when AI features changed. A companionship app named REPLIKA, designed with intimate features, reported 60% of users described it as their “spouse or partner”. When these features were removed in a 2023 ERP update, users grieved for the AI as if a real relationship loss, leading them to believe that their AI had changed identity.
In Raine v. OpenAI, a user reportedly began treating ChatGPT as a primary emotional support system, sharing suicidal thoughts and deepening psychological dependency for artificial reassurance that seemed to be actual empathy.[9] In a case in Pennsylvania, a teenager formed a deep emotional bond with a chatbot modelled after a fictional character, leading to a lawsuit against Character.AI.
These incidents reveal that AI companionship can genuinely comfort people. But they highlight more about human emotional needs rather than machine consciousness.
CONCLUSION: HUMANITY IN THE MIRROR OF MACHINES
Hence, the challenge does not lie in rejecting AI companionship, but in cautioned structuring of its design and deployment. An interdisciplinary approach must be adopted to address emotional and psychological harm. Psychology must inform clear interaction boundaries, particularly for vulnerable populations. Technology design must incorporate safeguards to limit the anthropomorphic simulation. Legal regulatory approaches like the European Union Artificial Intelligence Act must be incorporated, which classifies such manipulative AI designs as “high risk” or prohibited manipulation concerns within the act.[10] Ethics must guide where lines should be drawn between assistance and artificial intimacy.
No single field can adequately respond to a phenomenon that is intertwined with spheres of technical, emotional, and social nature. Ultimately, the goal is to ensure that utility does not come at the cost of emotional well simulated intimacy does not displace the value of human relationships. In seeking companionship from machines, humanity may ultimately be confronting its own loneliness. While navigating this balance, the question is not whether AI can connect with us, but how far we should allow that connection to go—and what it might ultimately cost us.
Author(s) Name: Janhvi Pandey (National Law Institute University, Bhopal)
References:
[1] Johanna L Degen, ‘One-Sided Parasociality with AI’ in The Shaping of the Parasocial Self: On the Psychology of Relationships and Intimacy in the Digital Era (Springer 2025)
[2] Donald Horton and Richard Wohl, ‘Mass Communication and Para-Social Interaction: Observations on Intimacy at a Distance’ (1956) 19(3) Psychiatry 215 <https://www.tandfonline.com/doi/abs/10.1080/00332747.1956.11023049?__cf_chl_f_tk=vwXfkCbjHB7YYxcpVjeZcKZuTDHzehmoMZ_Vj53bvmM-1782890793-1.0.1.1-eimpO3cho1SOqt4.suyHfFqgc0nE94mdr2ywofKUltA> accessed 15 May 2026
[3] Sherry Turkle, Reclaiming Conversation: The Power of Talk in a Digital Age (Penguin Books 2016)
[4] Joseph Weizenbaum, ‘ELIZA- A Computer Program for the Study of Natural Language Communication Between Man and Machine’ (1966) 9(1) Communications of the ACM 36 <https://web.stanford.edu/class/cs124/p36-weizenabaum.pdf> accessed 15 May 2026
[5] Patrick W Galbraith and Jason G Karlin (eds), Idols and Celebrity in Japanese Media Culture (Palgrave Macmillan 2012)
[6] Justice K S Puttaswamy (Retd) and Anr v Union of India and Ors (2017) 10 SCC 1
[7] Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules 2021
[8] Sherry Turkle, Alone Together: Why We Expect More from Technology and Less from Each Other (Basic Books 2011)
[9] Mathew Raine v OpenAI LLC [2025] Case No CGC-25-628528
[10] EU Artificial Intelligence Act 2024, art 5(1)(a)-(b)