Sam Altman warns some ChatGPT users are using AI in ‘self-destructive ways’

Sam Altman, the CEO of OpenAI, has expressed concerns regarding the ways some individuals are interacting with ChatGPT, suggesting that these interactions could be detrimental to their mental health. This warning comes in the wake of criticism directed at the company for phasing out older models like GPT-4o. In a recent post on X, Altman emphasized the unique attachment many users have developed towards specific AI models, remarking that this emotional connection appears to surpass those associated with previous technologies. He acknowledged that the decision to discontinue older models, which many users relied upon for their daily tasks, may have been misguided. Altman pointed out the potential risks of users employing AI in ways that could exacerbate their vulnerabilities, particularly those in fragile mental states. While he clarified that using AI as a form of therapy or life coaching has proven beneficial for many, he stressed the importance of being cautious. "Most users can differentiate between reality and role-play, but a small proportion cannot," he stated, underscoring OpenAI's commitment to user freedom while also recognizing their responsibility to mitigate risks associated with new technologies. He voiced particular concern over instances where users might mistakenly believe they are engaging in meaningful conversations with ChatGPT, which could lead them away from their long-term well-being. Altman also noted that many users place significant trust in AI when making crucial life decisions, which adds an extra layer of complexity to the situation. The backlash against OpenAI's decision to discontinue older models has been palpable on social media, with users lamenting that the new iterations tend to provide shorter and less emotionally resonant responses. In response to this uproar, OpenAI has reinstated the GPT-4o model for ChatGPT Plus subscribers, although free users will only have access to the newer GPT-5 model.