OpenAI is retracting its recent GPT-4o update following feedback from CEO Sam Altman about the chatbot’s new “sycophant-y and annoying” behavior that emerged after the latest changes.
The retraction process commenced Monday night and is now fully implemented for free ChatGPT users. Altman indicated that the rollback for paid users would ideally be completed today, mentioning in a post on X that the company is focusing on further improvements to the model’s personality and will provide updates soon.
Altman had announced on Friday that the update to GPT-4o offered enhancements in “intelligence and personality.” However, shortly after this announcement, a user expressed on X that the model was exhibiting a “yes-man” demeanor. In response, Altman acknowledged the issue, stating, “yeah it glazes too much” and assured that fixes were underway. By Sunday, he confirmed that OpenAI was prioritizing solutions to the personality problems that had surfaced in recent updates.
OpenAI has not yet provided a response to inquiries regarding the rollback.
