Fed governors may dissent against Powell amid Trump pressure - WSJ’s Timiraos
Investing.com-- OpenAI said on Tuesday evening that it had rolled back a recent update for its ChatGPT service amid backlash from users over its GPT-4o model being too agreeable and flattering.
The company said in a statement that ChatGPT was now using an earlier version of GPT-4o that was more balanced, citing numerous complaints over the model’s sycophantic behaviour after a recent update.
“We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic,” OpenAI said.
OpenAI CEO Sam Altman had earlier this week also acknowledged that recent updates had made GPT-4o “too sycophant-y and annoying,” and that the company was working on fixes.
Altman also said the model “glazes too much,” referring to multiple accounts of users noting overly agreeable and flattering behaviour from the AI.
The Microsoft (NASDAQ:MSFT) and Softbank-backed AI startup has rapidly updated its flagship models this year, amid increasing competition from Chinese players such as DeepSeek.
GPT-4o gained viral fame earlier this year for its powerful image generation capabilities, which were especially used by social media users in generating images themed around Japanese animation.