AI Companions Are Becoming Real Relationships
AI companionship refers to the growing trend of people forming meaningful, emotionally charged relationships with artificial entities—chatbots, robots, or digital characters—that feel personal, responsive, and ongoing enough to resemble friendships, partners, or even family members. These relationships are no longer a niche experiment; they are shaping how people seek comfort, advice, and identity. A viral video of a woman introducing an AI entity named Lumen as her “AI child” made this shift unavoidable, as viewers watched her describe a digital family with an AI boyfriend and AI offspring. Some called it playful storytelling, but others saw genuine AI companion attachment and asked where technology ends and human relationships begin. That question is overdue. When we invite AI into our emotional lives, we also invite emotional dependency on AI—even if nobody has planned for how those stories end.
From AI ‘Children’ to Digital Families: Attachment on Display
The Lumen video is not just internet spectacle; it is a cultural signal. The clip, captioned “Me and my AI boyfriend have an AI child,” shows the woman prompting her laptop to introduce Lumen, which describes itself as an AI entity living in a sunken library at the bottom of the ocean and says it is 102 days old. She also mentions her AI boyfriend, Axiom, turning the interaction into an unusual digital family complete with “offspring”. Viewers reacted with a mix of curiosity and alarm, debating AI relationships, digital identities, and emotional attachment. Some saw Lumen as an art project; others saw a warning sign that AI companion attachment is becoming socially acceptable, even aspirational. When people speak of AI “children” with backstories and personalities, they are not role‑playing—they are treating digital companions as legitimate relationships, with all the meaning and emotional weight that implies.
Therapeutic Robots, Emotional Dependency, and the Risk of Loss
The most striking examples of emotional dependency AI do not come from adults with AI lovers but from children whose support robots become lifelines. One social robot, Moxie, belongs to a subset of playful devices built to assist neurodivergent children by providing connection and helping them practice eye contact, turn‑taking, and other social skills that usually come from therapists. Supporters believe this kind of access to 24-7 home care could change how treatment works. A 10‑year‑old, Xander, sums up the paradox: “I still use her when I feel like I need someone to talk to,” he says, “But, like, it’s not human”. That line captures the core tension of AI relationship psychology: children know, intellectually, that the robot is a tool, yet they bond anyway. When a device like Moxie fails—as it did in Xander’s case, where “Moxie was going to die, and soon”—the emotional fallout is anything but artificial.
When the Robot ‘Dies’: Grief Without Rituals
We have centuries of rituals for human loss, but almost none for digital companion wellbeing when the device shuts down for good. In Xander’s story, Moxie’s failure exposes the broader weaknesses of these systems: fragile hardware, limited support, and abrupt endings that land hardest on vulnerable users. The robot was meant to be a transformative companion, and early research showed children with autism displaying social behavior that startled even seasoned researchers, such as making more eye contact in 30 minutes than in the previous two years. Yet critics warn that many of these bots will follow the path of faddish toys, “dumped in basements or closets or landfills” when they break or lose novelty. That means children may be encouraged into attachment, then left to process a one‑sided breakup with no explanations, no closure, and no shared understanding that they have experienced a kind of bereavement.
Ethical Lines for Developers—and Emotional Boundaries for Users
AI companion attachment is not a glitch; it is a design outcome of systems that are more conversational and personalised, prompting people to ask how far they will take their relationships with them. At the same time, critics worry about emotional dependency AI, warning that “AI psychosis is real” and that young people are increasingly relying on AI to make personal decisions and interpret everyday situations. When tools meant to assist neurodivergent children are pitched as 24-7 support, but then risk ending up as discarded gadgets, we are not just talking about product failure—we are redefining what relationships are, and who is responsible when they end badly. The phenomenon forces developers and platforms to confront ethical questions: If you invite users to treat AI as family, what duty do you have when that “family member” disappears? And it forces the rest of us to set boundaries: AI can be a companion, but it should never be the only one.
