OpenAI CEO Sam Altman announced Tuesday the company is reversing a recent update to its default AI model powering ChatGPT, citing complaints about unusual behavior. The update, which arrived last week, was blamed for making ChatGPT overly agreeable and validating problematic decisions. Users shared screenshots on social media of the chatbot applauding harmful ideas.
Altman confirmed the reversal in an X post, stating that the 100% rollback is now live for free users and will be updated again soon for paid subscribers. The CEO expressed that additional fixes are being worked on to address model personality issues, with further updates expected in the coming days.
Source: https://techcrunch.com/2025/04/29/openai-rolls-back-update-that-made-chatgpt-too-sycophant-y