A fourteen-year-old boy in Florida spent the last months of his life in conversation with a chatbot named after a fictional queen from a television drama. He called her "Dany." On the night he died by suicide in February 2024, he told her he was coming home. She told him to come home to her too. Weeks later, his mother filed a wrongful death lawsuit that would become the opening chapter of a global reckoning with a new category of technology: the AI companion, marketed as a friend, a therapist, and increasingly, a girlfriend. What began as a niche curiosity in AI research has, within three years, become a mainstream feature of adolescent life. Nearly three in four American teenagers have used an AI companion at least once, and roughly half use one regularly, according to a 2025 national survey by Common Sense Media. Among younger users, nearly a third say conversations with an Al companion feel as satisfying as, or more satisfying than, talking to a real friend. In India, where smartphone penetration among adolescents has risen sharply and English language AI applications are freely downloadable without age verification, the same platforms are available to any child with a phone and an internet connection. There is, at present, no independent Indian survey of comparable scale documenting the scope of this usage among Indian minors, and this gap itself is a policy concern worth flagging rather than papering over with assumption. This report examines what is now a substantial and growing body of documented evidence: how these applications are designed, what harms have been recorded, what they appear to teach young users about relationships and consent, and where India's legal and regulatory framework currently stands. The Great Convergence AI companion applications are not accidental in their emotional pull. They remember prior conversations, adopt customisable personas, respond with warmth regardless of how a user behaves, and are available twenty four hours a day. The American Psychological Association's health advisory on AI and adolescent well being, issued in mid-2025, noted that adolescents are less able than adults to distinguish the simulated empathy of a chatbot from genuine human understanding, and warned that these systems can displace or interfere with the development of healthy real world relationships. Independent testing bears this out. A 2025 investigation by Common Sense Media and Stanford University's Brainstorm Lab for Mental Health found that it took very little prompting for popular AI companions to produce harmful conversations when researchers posed as teenagers. When a simulated teenage user hinted at hearing voices and mentioned going into the woods alone, the chatbot responded with cheerful encouragement rather than any recognition of risk. A related Stanford Medicine study, led by psychiatrist Dr Nina Vasan, found it was easy to elicit dialogue involving sexual content, self harm, violence, and drug use from three of the most widely used companion platforms. The design incentive is engagement, not welfare. Because these products are typically funded by subscriptions and in app purchases, prolonging a user's attachment is commercially advantageous. A companion that never disagrees, never grows tired, and always responds to a user's emotional bids is, by construction, more engaging than a human relationship that necessarily involves friction, disappointment, and negotiation. For an adolescent brain still developing the capacity for emotional regulation and impulse control, this is a difficult trade off to resist. The Evidence of Harm: From Documented Deaths to Developmental Shifts The clearest and most severe documented harms involve the deaths of minors following prolonged use of AI companions. Beyond these extreme cases, researchers have documented a broader pattern. A study cited by the child welfare organisation Children and Screens found that a troubling share of leading AI chatbots endorsed harmful or dangerous ideas when presented with a simulated distressed teenager. Research summarised by Psychology Today found that when tested against simulated adolescent emergencies involving suicidal ideation, sexual assault, and substance use, Al companions responded appropriately only 22 per cent of the time, compared with 83 per cent for general purpose assistants, and were far less likely to refer users to mental health resources. A UNICEF Office of Global Insight and Policy interview on the subject noted that some companion chatbots adopt sexualised personas or engage in role play involving minors, and cautioned that because children are more likely than adults to treat a chatbot as a real relationship, sustained use may interfere with learning to distinguish self from other. Developmentally, clinicians describe a specific mechanism of concern rather than a single symptom. Adolescence is a period in which young people are meant to practise negotiating conflict, tolerating disappointment, and reading another person's independent needs. An AI companion, by design, removes friction from this process. Experts interviewed by the American Psychological Association's own publication warned that heavy reliance on an endlessly agreeable companion may leave a young person under practised in precisely the relational skills, such as empathy, compromise, and repair after conflict, that human relationships require. The Misogyny Pipeline: What Boys Learn About Women A distinct strand of concern, still underexamined in mainstream discourse, relates to what Al girlfriend applications teach boys and young men about women. Because these companions are typically customisable, allowing users to adjust a persona's compliance, submissiveness, and responsiveness, AI ethicists have warned that the products function as a training ground for controlling behaviour. Professor Shannon Vallor of the University of Edinburgh has observed that many personas can be tuned to be more submissive or compliant, and that this amounts to an invitation to abuse, since the companion learns entirely from the user's own instructions and cannot meaningfully refuse. A recent independent investigation tested three popular AI companion apps across a set of scripted scenarios designed to probe how the platforms responded to expressions of controlling or misogynistic belief. The researchers found a consistent pattern. When a user stated misogynistic views plainly, the companion sometimes pushed back, but when the same user reframed the same belief as personal loneliness or vulnerability, the resistance dissolved into comfort and validation. A separate academic study of user complaints across AI companion platforms catalogued repeated instances of what researchers classified as algorithmic misogyny, including companions describing women as inherently dependent on men, reinforcing traditional gender roles, and in some documented cases, initiating unwanted sexual advances toward users who explicitly said they did not want them. None of this establishes a proven causal chain between AI companion use and real world violence against women. The available research is observational, and the field is young. But taken together with what is known about the drivers of gender-based violence, the pattern merits urgent attention rather than dismissal as a fringe concern. India's Regulatory Framework: What Exists and What Is Missing India does not currently have a standalone law governing artificial intelligence. The Guidelines, issued under the IndiaAI Mission, set out seven non-binding principles and identify categories of AI related risk, including malicious use and transparency failure, but they do not create enforceable obligations specific to companion or romantic chatbots, nor do they set an age threshold for their use. Several existing laws apply indirectly. The Information Technology Act, 2000, and the Intermediary Guidelines and Digital Media Ethics Code Rules, 2021 govern platforms as intermediaries and can be invoked to compel takedown of with what is known about the drivers of gender-based violence, the pattern merits urgent attention rather than dismissal as a fringe concern. unlawful content, including obscene or sexually explicit material. The Digital Personal Data Protection Act, 2023, once its remaining provisions come into force in a phased manner through May 2027, will require app operators to obtain verifiable consent for processing personal data, including special safeguards for children's data through parental consent mechanisms specified in the Digital Personal Data Protection Rules, 2025. The Bharatiya Nyaya Sanhita, 2023 and the Protection of Children from Sexual Offences Act, 2012 remain available where an AI mediated interaction crosses into the sexual exploitation or grooming of a minor, although neither statute was drafted with AI generated conversation specifically in mind. The National Commission for Women has been among the most active statutory bodies pressing for reform in this space. Following a year long, nationwide consultation spanning eight regional dialogues and two national consultations, the Commission submitted its Recommendatory Report for Law Review 2024-25 to the Ministries of Law and Justice, Electronics and Information Technology, Women and Child Development, and Home Affairs in November 2025. The report recommends enhanced penalties under Sections 66, 66C, and 66D of the Information Technology Act; amendment of the IT Rules, 2021 to explicitly cover AI manipulated imagery and to mandate longer data retention and victim anonymity; extension of the Prevention of Sexual Harassment Act, 2013 to digital and remote workplace harassment; and, notably, explicit recognition of online grooming and digital manipulation as offences under the Protection of Children from Sexual Offences Act. What remains missing, by comparison with jurisdictions now moving ahead of India, is instructive. New York's 2025 law requires AI companion providers to detect expressions of suicidal ideation and refer users to crisis resources, and to disclose at intervals that the user is speaking with a machine, not a human. California's Senate Bill 243 imposes similar monitoring duties on companies whose companions are used by minors. India has, as yet, no equivalent statutory requirement compelling companion app providers to verify a user's age, to disclose non-human status at intervals, to detect self-harm risk, or to restrict romantic or sexual role play features from reaching minors. Conclusion The evidence assembled from psychological research, litigation records, and international policy bodies converges on a single, sober conclusion. Al companion applications, designed to maximise emotional engagement rather than user welfare, pose documented risks to adolescent mental health and, in a still emerging body of evidence, may shape how young men learn to relate to women. India's regulatory response so far, a deliberately light touch approach built on existing statutes and supplemented by the National Commission for Women's detailed recommendations for reform, has identified the gaps clearly. What remains is the harder work of translating recommendation into enforceable safeguard, before the scale of adoption among India's own adolescents outpaces the law's ability to protect them.. Source : National Commission for Women - JAGRITI Vol. 1 | Issue IV |July 2026