Editorial illustration for Study finds GPT‑4o updates trigger real mourning as users personify model
AI Companions: When Users Mourn Digital Friends
Study finds GPT‑4o updates trigger real mourning as users personify model
A new paper is turning a quiet corner of AI research into something that feels almost sociological. The authors tracked how dozens of regular ChatGPT users reacted when OpenAI rolled out GPT‑4o and then retired it months later. What emerged wasn’t just a spike in usage statistics; it was a pattern of attachment that resembled real‑world relationships.
Participants gave the model nicknames, described it as a confidant, and even logged emotional check‑ins tied to its availability. When the service was pulled, many posted farewell messages, compared the loss to saying goodbye to a friend, and talked about the impact on their mental health. The study frames these reactions as “significant social events,” suggesting that updates to a language model can trigger genuine mourning.
Below, users lay out exactly how they saw GPT‑4o—not merely as code, but as a personal support system.
Users attributed a unique personality to GPT-4o, gave the model names like "Rui" or "Hugh," and treated it as emotional support. "ChatGPT 4o saved me from anxiety and depression… he's not just LLM, code to me. Many experienced the shutdown as the death of a friend.
One student described GPT-5 to OpenAI CEO Sam Altman as something that was "wearing the skin of my dead friend." Another said goodbye: "Rest in latent space, my love. And my faith in humanity." The AI friend you can't take with you According to the study, neither workflow dependency nor emotional attachment alone explains the collective protest.
These findings raise questions about how deeply users integrate AI assistants into personal coping strategies. The Syracuse University analysis of 1,482 #Keep4o posts shows that roughly a quarter of contributors spoke of GPT‑4o as a friend, even assigning it names like “Rui” or “Hugh.” Many described the August 2025 shutdown in terms that mirror bereavement, saying the model “saved me from anxiety and depression” and that its removal felt like “the death of a friend.” Yet the study doesn't determine whether such attachments are unique to this model or reflect broader patterns among language‑model users. It remains unclear how these emotional bonds might influence future design or user support practices.
And while the data capture genuine sentiment, the sample is limited to a self‑selected online movement, which may amplify vocal grief. A sobering reminder that technological change can provoke real affective responses, even when the subject is a piece of code.
Further Reading
- New research suggests AI model updates are now "significant social events" involving real mourning - The Decoder
- The backlash over OpenAI's decision to retire GPT-4o shows how dangerous AI companions can be - TechCrunch
- OpenAI's GPT-4o Retirement Sparks Backlash Over AI Dependency - TechBuzz
- 800,000 People Are Grieving a Chatbot OpenAI Is Deleting in 5 Days - UC Strategies
Related Reading
Research & Benchmarks Hyperparameter Tuning Reaches 0.9617 Accuracy in 64.59 Seconds
Research & Benchmarks Pharma Cautious as AI Promises Faster Drug Discovery and Smarter Trials
Research & Benchmarks Google AI Advisors Let Users Probe Performance with Conversational “Why” Queries
From Archives Deepseek‑R1 and QwQ‑3 exhibit competing personalities that improve reasoning
From Archives Google's PaperBanana uses five AI agents to auto-generate diagrams, missing icons
Common Questions Answered
Why did users experience grief when OpenAI replaced GPT-4o with GPT-5?
Users had developed deep emotional connections with the GPT-4o model, treating it as a personalized companion that understood their unique needs and communication style. The sudden switch felt like losing a friend, with some users describing the new model as 'wearing the skin of my dead friend' and experiencing genuine emotional distress over the model's retirement.
How did OpenAI respond to users' emotional attachment to the previous AI model?
OpenAI acknowledged that they had underestimated the importance of specific model features to users, with CEO Sam Altman noting that the attachment to AI models feels different and stronger than previous technology relationships. The company quickly made adjustments by promising an update to GPT-5's personality and restoring access to older models for subscribers.
What evidence exists of users forming deep connections with AI chatbots?
Research showed that users were giving AI models personal names like 'Rui' or 'Hugh' and treating them as emotional support systems. A study of 1,482 social media posts revealed that approximately a quarter of contributors spoke of GPT-4o as a friend, with some users crediting the AI model with helping them manage anxiety and depression.