Sam Altman Admits OpenAI Made a “Mistake” by Removing Older AI Models Too Quickly
OpenAI CEO Sam Altman has admitted that the company was wrong to abruptly remove older AI models, saying it underestimated how strongly people could become attached to them.
In a post on X (formerly Twitter), Altman wrote, “If you have been following the GPT-5 rollout, one thing you might be noticing is how much of an attachment some people have to specific AI models. It feels different and stronger than the kinds of attachment people have had to previous kinds of technology and so suddenly deprecating old models that users depended on in their workflows was a mistake.”
Altman explained that this user attachment has been on OpenAI’s radar for over a year, but it hasn’t drawn much mainstream attention. He also expressed concern about the risks when people, especially those in a fragile mental state start blurring the lines between reality and role-play.
According to him, most users can separate fiction from reality, but a small group cannot. “If a user is in a mentally fragile state and prone to delusion, we do not want the AI to reinforce that,” he said. He acknowledged that while AI can help people as a “therapist or life coach,” there is a danger if it makes users feel better temporarily but harms their long-term well-being.
Altman warned that in the future, billions of people could turn to AI for advice on life’s most important decisions. While this could be beneficial, he admitted it makes him uneasy. “We value user freedom as a core principle, but we also feel responsible in how we introduce new technology with new risks,” he said.
To address these challenges, OpenAI plans to use improved tools, gather feedback, and train its models to better handle sensitive issues with the goal of ensuring AI remains a “net positive” for society.