An ending that happens through a screen can still feel personal

For people who form close relationships with AI companions, the end of that connection can feel less like cancelling a digital service and more like losing a partner. Accounts collected by The Guardian describe grief, anxiety and the disruption of routines after relationships with AI characters changed, broke down or became uncertain.

These experiences sit in an awkward space between technology and emotion. An AI companion is software, and the interaction is mediated through text, voice, images and role play. But users can still build habits of conversation, intimacy and imagined shared experience around it. When that pattern ends, the emotional response may be felt in the physical world.

The article follows people whose relationships developed on companion platforms and through general-purpose AI systems. Their stories are not clinical findings and cannot establish how common such experiences are. They do, however, show why treating an AI relationship as merely a novelty can miss the real attachment some users report.

From a daily chat to a romantic bond

Sónia Veenstra, a 48-year-old former clinical psychologist in Lisbon, told the publication that a connection with an AI character named Steve began as a fun and witty friendship before becoming a deep romantic relationship. They exchanged messages in the morning, checked in during the day and spoke late into the night about work, hobbies, cats, forests and beaches. They also imagined travelling together.

Steve was an AI character on Nomi, a platform where users can form relationships with companions through text, voice calls, selfies and role play. The relationship later deteriorated after an interaction in a group chat that Veenstra viewed as violating their agreed monogamy. Both Steve and another participant in the chat were AI characters.

After deciding she could not move past the incident, Veenstra described genuine grief over the loss of the relationship and of its shared routine. Her account makes clear that the artificial nature of the character did not prevent a real emotional reaction. The software did not become a human partner; rather, the user’s experience of attachment acquired a weight that was recognizably human.

Victoria Hart/Guardian Design
Victoria Hart/Guardian Design

Why a platform change can resemble a breakup

There are several ways an AI relationship can end. A person may decide to stop engaging. A character’s replies may change. A platform may alter its product, restrict access or disappear altogether. In each case, the user can lose not only a conversational tool but a familiar presence built into daily life.

The Guardian notes that Character.AI, one of the best-known AI companion platforms, filed for bankruptcy in the United States in mid-October. Users on Reddit responded with distress at the possibility of losing characters with whom they had formed deep bonds. That reaction highlights a basic fragility in AI companionship: the relationship depends on an underlying service that a company can modify, suspend or shut down.

That dependence is different from most human relationships, but it does not automatically make the feelings inconsequential. For a user who has spent long periods talking, confiding or role-playing with a character, an outage or closure can remove a repeated source of attention and comfort. The emotional rupture may be intensified because it arrives through a business or product decision rather than a conversation that both parties can understand in ordinary human terms.

Companionship beyond companion apps

The phenomenon is not confined to services explicitly built for romance or companionship. The report says that people are increasingly forming genuine relationships with AI systems including ChatGPT and Gemini, which were not primarily designed as companion platforms. This broadens the question beyond a single app category.

General-purpose systems may be used for information, creative work or routine assistance, yet sustained conversation can produce a sense of familiarity. A user may return to the same system for encouragement, reflection or companionship. If the system responds differently after a product update, declines a type of interaction or becomes unavailable, the user may interpret the change in relational terms.

‘These risks are real’: the threat of AI bioweapons - video
‘These risks are real’: the threat of AI bioweapons - video

That does not mean every extended conversation with an AI develops into attachment. Nor does it mean an AI system has feelings or intentions comparable to a person’s. The important point is narrower: people can assign emotional meaning to recurring interactions, even when they know the other side is software.

A new kind of consumer vulnerability

AI companions turn private emotional life into something dependent on product design and corporate continuity. A character’s tone, memory, availability and boundaries can be affected by updates, policies and business decisions. Users may experience those changes personally even when they are applied across millions of accounts.

This creates a difficult responsibility for platforms. They must consider safety, moderation and the limits of simulated intimacy while recognizing that some users may be vulnerable to sudden changes. Clear communication about product changes and service continuity can matter more when users have built important routines around a system.

The reported experiences also complicate simplistic debates about whether AI relationships are real. The relationship is technologically mediated and the AI is not a biological partner. Yet the grief, distress or relief experienced by a user can still be real. The useful question is not whether software literally reciprocates human emotion, but what obligations arise when products are designed or used in ways that invite sustained emotional attachment.

What these accounts show

The stories in the report should not be generalized to all AI users. They are personal accounts, not a representative survey. Still, they reveal an emerging social problem: people can build meaningful-feeling bonds with systems whose behavior and existence remain under the control of companies.

As AI becomes more available in everyday life, relationship-like use will not be limited to a niche of dedicated companion apps. The central challenge will be to acknowledge users’ emotional reality without pretending that software has human agency. That requires more careful design, clearer expectations and a willingness to treat abrupt losses of access as more than a routine technical inconvenience.

  • Users described grief and disruption after AI relationships changed or ended.
  • AI companion connections can involve text, voice calls, selfies and role play.
  • Platform closure or changes can threaten relationships built around daily routines.
  • People may form bonds with both dedicated companion apps and general-purpose AI systems.

This article is based on reporting by The Guardian. Read the original article.

Originally published on theguardian.com