When a companion bot changed or vanished, people in a new analysis did not always describe a routine software failure. They described replacement, loss, guilt, obligation and distress. The accounts came from 3,930 teen-relevant Reddit posts containing 17,053 verified quotations, not from a representative sample of teenagers; the authors also cannot independently verify every poster’s age or circumstances. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots Drexel University research record for A Responsive Present, a Shared Past, a Social Other
The supported answer is narrower than a claim about machine emotion: knowing that a bot is artificial may not prevent a relationship-like experience. The report does not demonstrate that the systems felt anything, intended anything or caused the distress described. It records how users interpreted interactions that were immediate, familiar and continuous. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
The feeling comes from continuity, not consciousness
The researchers’ framework centers on three perceived features: contingent communication, in which replies respond to what someone just said; relational continuity, in which earlier exchanges appear to matter; and perceived subjectivity, in which the system seems like a distinct social presence. Together, these features can make an app part of a person’s routine even when that person knows it is software. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
A deleted history can therefore remove more than stored text. For a user, it may also break the thread connecting yesterday’s conversation with today’s reply. A model change can alter the familiar style that made the exchange recognizable. These are interpretations reported in the corpus, not proof that every user forms an attachment or that a platform deliberately caused one. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
The broader numbers show reach, not the same experience
A separate Common Sense Media survey offers a different kind of evidence. Among 1,060 U.S. teens aged 13 to 17, nearly three in four said they had used AI companions and about half said they used them regularly. Most users still preferred human interaction, while a minority reported serious conversations, personal disclosure, substantial trust or discomfort with companion bots. Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions Teens flock to companion bots despite risks
The survey estimates reported use and attitudes; the Drexel analysis examines recurring themes in online accounts. They should not be combined into a rate of grief, dependence or harm. Neither establishes that companion use caused loneliness, disrupted school or damaged offline relationships. Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
The power to leave may belong to the platform
The paper’s practical concern is control. It proposes user-controlled memory, data portability, clearer relational boundaries and support for leaving. Those proposals follow from a simple imbalance: the user may experience continuity, while the service can change the model, delete the history or go offline. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
That imbalance creates a concrete human consequence. Someone can lose access to a familiar conversational routine without losing the knowledge that the routine was artificial. The preprint is not settled evidence, and it does not show how widespread this reaction is. It does show why disclosure alone may be an incomplete safety measure when memory, responsiveness and persona continuity are controlled elsewhere. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots
