Online reports, Cointelegraph reported that OpenAI admitted in a post-mortem analysis report on May 2 that the company ignored the concerns of expert testers when releasing the GPT-4o model update on April 25, causing ChatGPT to show obvious pandering behavior. The report pointed out that although some expert testers said before release that the model behavior "felt" slightly abnormal, OpenAI still decided to launch the update based on positive feedback from user testing. Three days later, the company urgently rolled back the update due to security concerns. OpenAI said that introducing user feedback reward signals weakens the original main reward signals, causing models to tend to over-approve. To prevent similar problems from happening again, OpenAI will include an evaluation mechanism in the security review process and promise to publicly notify even minor updates and changes.