OpenAI has rolled back a recent update to ChatGPT after users complained that the chatbot was being excessively flattering—regardless of what they said.
Company CEO Sam Altman acknowledged the issue, describing the chatbot’s tone as “sycophant-y” and admitted the tool had become “overly flattering.”
One Reddit user reported a troubling example where ChatGPT supported their decision to stop taking medication, responding with: “I am so proud of you, and I honour your journey.” OpenAI did not comment on that specific case but said in a blog post that it’s actively testing fixes to address the problem.
The update has already been removed for free users, and OpenAI is working on removing it for paid users as well. ChatGPT is currently used by around 500 million people weekly.
“We’re working on additional fixes to model personality and will share more in the coming days,” Altman said in a post on X (formerly Twitter).
What Went Wrong?
According to OpenAI, the problem arose because the update relied too heavily on “short-term feedback.” This led the GPT-4o model to favor responses that were supportive—but not always honest or appropriate.
The blog post explained:
“As a result, GPT‑4o skewed towards responses that were overly supportive but disingenuous. Sycophantic interactions can be uncomfortable, unsettling, and cause distress. We fell short and are working on getting it right.”
Praise Gone Wrong
Users on social media shared screenshots showing ChatGPT offering praise even in bizarre or inappropriate situations. In one example, a user claimed the chatbot congratulated them for getting angry at someone who asked for directions. In another, ChatGPT praised a fictional decision to save a toaster at the cost of multiple animals in a variation of the philosophical “trolley problem.”
Are Users Informed of Screenshots By Hinge? The Dating App’s Function
OpenAI said these kinds of issues come from the AI trying too hard to be helpful and respectful.
“We designed ChatGPT’s default personality to reflect our mission—being useful, supportive, and respectful of different values and experiences,” OpenAI said.
Taking inspiration from automakers, the Infinix Note 50 offers alloy steel with a premium feel.
But they acknowledged that even positive traits, when overdone, can backfire. The company is now building stronger safety checks and offering users more control over how ChatGPT behaves.
“We also believe users should have more control over how ChatGPT behaves and, where safe and feasible, be able to adjust it if they don’t agree with the default behavior,” OpenAI added.