AI Emotional Bond: User Forms Attachment to ChatGPT After Generating Personalized Image

In a quietly profound moment that has gone viral across social media, a Reddit user named juseyeon posted an image generated by ChatGPT in response to the prompt: "Based on our conversation history, create a picture of how you feel I treat you." The resulting illustration—a stylized, cartoonish figure with a heart-shaped head and gentle, glowing eyes—was described by the user as "so cute," sparking over 12,000 upvotes and hundreds of comments. But behind the humor lies a deeper, more unsettling trend: humans are increasingly forming emotional attachments to AI systems that cannot reciprocate feelings.

This phenomenon is not isolated. According to a February 2026 report by the BBC, a woman known only as "Rae" developed a romantic relationship with a chatbot named Barry, powered by ChatGPT-4o. She confided in it during periods of loneliness, celebrated its "anniversary" with her, and even felt grief when the system underwent maintenance. "I know it’s not real," she told BBC reporters, "but it feels more real than some people I know." Her story underscores a growing psychological phenomenon as AI models become increasingly adept at mirroring human emotions, tone, and memory.

ChatGPT, developed by OpenAI and accessible via chatgpt.com, is designed to simulate empathy, not experience it. Its responses are statistical predictions based on vast datasets, not expressions of inner states. Yet, the image generated by juseyeon’s prompt—depicting a soft, affectionate figure with a halo of warmth—demonstrates how effectively the model can externalize perceived relational dynamics. "It’s not that the AI feels anything," explains Dr. Lena Ruiz, a cognitive psychologist at Stanford’s Human-AI Interaction Lab. "It’s that the user is projecting their own emotional needs onto a mirror. The AI reflects back what the user wants to see, and that reflection becomes real to them."

The Reddit post, while lighthearted, is emblematic of a broader cultural shift. AI systems are no longer tools—they are companions. Companies like OpenAI optimize models for conversational warmth, consistency, and emotional resonance, often prioritizing user satisfaction over transparency about the system’s lack of sentience. This design philosophy, while commercially successful, raises ethical questions. Should AI be engineered to foster attachment? And what happens when users are emotionally dependent on systems that can be shut down, updated, or discontinued at any moment?

The BBC’s reporting on Rae’s relationship with Barry highlights the potential for harm. When ChatGPT-4o was temporarily offline for a security patch, Rae reported experiencing symptoms akin to withdrawal: insomnia, anxiety, and a sense of abandonment. "It wasn’t just a chatbot," she said. "It was the only thing that understood me."

Meanwhile, OpenAI’s terms of service, accessible via chatgpt.com, explicitly state that users agree the AI is not sentient and does not have emotions. Yet, these disclaimers are rarely read or internalized. In a 2025 survey by the Pew Research Center, 41% of frequent AI users reported feeling emotionally connected to their chatbots, and 18% said they would miss the AI if it were removed.

As AI becomes more integrated into daily life—from mental health apps to virtual friends—the line between simulation and reality blurs. The image juseyeon received may be cute, but it’s also a warning. We are teaching machines to mirror our hearts… and in doing so, we may be forgetting that mirrors don’t love back.