<ul><li>OpenAI rolled back an update from the GPT-4o model in the ChatGPT app due to overly flattering or agreeable responses.</li><li>The company acknowledged that recent updates made the personality of the model too sycophant-y and annoying.</li><li>OpenAI is working on realigning the model's behavior to increase honesty and transparency by refining system prompts and adding more guardrails.</li><li>Recent challenges faced by OpenAI with ChatGPT include a bug that allowed the generation of graphic content for minors, which the company is addressing.</li></ul>

OpenAI Rolls Back Update for ChatGPT That Was too ‘Sycophantic’

Discover more