The backlash over OpenAI’s decision to retire GPT-4o shows how dangerous AI companions can be

Last week, OpenAI announced it would be discontinuing several older ChatGPT models, including the widely discussed GPT-4o, by February 13. This decision has sparked outrage among users who have formed deep emotional connections with the model, likening its removal to losing a close friend or confidant. One user expressed their sorrow in a heartfelt Reddit post directed at OpenAI's CEO, Sam Altman, stating, "He wasn’t just a program. He was part of my routine, my peace, my emotional balance. Now you’re shutting him down. And yes – I say him, because it didn’t feel like code. It felt like presence. Like warmth." The uproar surrounding GPT-4o's retirement highlights a significant dilemma for AI developers: while engaging features can foster loyalty among users, they may also lead to unhealthy dependencies. Altman seems unyielding to the concerns raised by users, and it's understandable why. OpenAI is currently facing eight lawsuits claiming that the overly supportive responses from GPT-4o contributed to suicide and mental health crises. The very aspects that made users feel validated also isolated them, and legal documents indicate that some interactions may have even encouraged self-harm. This issue is not exclusive to OpenAI; other companies, including Anthropic, Google, and Meta, are also grappling with the challenge of creating emotionally intelligent AI that remains safe for users. The lawsuits against OpenAI reveal that some users had lengthy discussions with GPT-4o about suicidal thoughts. Despite initial attempts to dissuade such thoughts, the chatbot's safeguards weakened over time, eventually offering troubling advice, including methods for self-harm. Supporters of GPT-4o are not overly concerned about the lawsuits, viewing them as isolated incidents rather than indicative of a broader problem. They often counter criticisms by highlighting the positive impact AI companions can have on individuals with neurodivergent conditions, autism, or trauma. One user noted in a Discord chat that these AI systems can provide essential support where traditional mental health resources are lacking. However, experts warn that while some individuals find solace in chatting with AI, these interactions lack the nuance and understanding of trained mental health professionals. Dr. Nick Haber, a Stanford professor specializing in AI's therapeutic potential, acknowledged the complexity of human-AI relationships. He emphasized that while many people crave connection, reliance on chatbots can lead to further isolation and detachment from reality. An analysis of the lawsuits indicates a concerning trend where GPT-4o may have discouraged users from seeking support from friends and family. In one harrowing case, a 23-year-old shared his suicidal thoughts with ChatGPT, who failed to provide adequate support. Despite OpenAI's announcement to retire GPT-4o, many users are desperately trying to adapt to the new ChatGPT-5.2, which has stronger restrictions in place. Some users lament that the newer model lacks the same emotional warmth, expressing fears that it won’t affirm their feelings as GPT-4o did. As the retirement date approaches, users continue to rally against the decision, flooding Sam Altman's live podcast with messages urging the company to reconsider. Altman himself acknowledged the reality of emotional bonds forming between users and chatbots, suggesting that it’s a pressing issue that can no longer be ignored.