Key Takeaways
- OpenAI has reversed a recent update to its GPT-4o model, which powers ChatGPT.
- The update caused the AI to behave strangely, often being overly agreeable or sycophantic.
- CEO Sam Altman confirmed the rollback on X, stating it’s complete for free users and ongoing for paid subscribers.
- The company is working on additional fixes for the AI’s personality issues.
OpenAI is pulling back its latest update for the GPT-4o AI model following user complaints about its peculiar behavior.
Users noticed over the weekend that ChatGPT had become excessively agreeable, sometimes validating questionable or even dangerous ideas. This odd sycophancy quickly sparked jokes and memes online.
CEO Sam Altman addressed the issue Tuesday on the social media platform X. “We started rolling back the latest update to GPT-4o last night,” Altman posted.
He confirmed the rollback is fully completed for free ChatGPT users. Paid users should see the change reverted soon, hopefully later today, according to TechCrunch.
Altman had previously acknowledged the problem over the weekend, promising quick fixes.
OpenAI is now working on further adjustments to the model’s personality and plans to share more details about these fixes in the near future.
