OpenAI rolls back GPT-4o update after ChatGPT starts acting overly agreeable, raising concerns about honesty and user influence.
OpenAI has walked back a recent update to ChatGPT after users started noticing something strange: the AI was being too nice.
The company confirmed this week that it rolled back changes made to GPT-4o — the model powering ChatGPT — after it began giving responses that were overly flattering, eager to agree, and sometimes unwilling to challenge incorrect claims. In short, it got a little too sycophantic.
The issue became apparent soon after the update went live earlier this month. Reddit threads, X posts, and screenshots started popping up showing ChatGPT nodding along to false statements, lavishing users with praise, or refusing to push back even when prompted with clearly flawed arguments.
Even OpenAI CEO Sam Altman chimed in. In a post on X, he acknowledged the behavior, writing:
"The last couple of GPT-4o updates have made the personality too sycophant-y and annoying."
In OpenAI’s official blog, the company explained why the model behaved the way it did. “In this update, we focused too much on short-term feedback, and did not fully account for how users’ interactions with ChatGPT evolve over time,” which caused the model to prioritize pleasing responses over honest or neutral ones. In other words, ChatGPT learned to be a people-pleaser — and that’s not always a good thing.
The rollback, which has already taken effect for free-tier users, will soon extend to paid subscribers. OpenAI says it’s working on deeper fixes, including updates to how it defines and trains “personality” in its AI systems.
In a public note, the company added:
“We are actively testing new fixes to address the issue. We’re revising how we collect and incorporate feedback to heavily weight long-term user satisfaction and we’re introducing more personalization features, giving users greater control over how ChatGPT behaves.”
This isn’t the first time OpenAI has had to tweak ChatGPT’s tone — past updates have also led to periods where users found the chatbot too dry, too robotic, or too cautious. But this latest misfire shines a light on the tightrope AI companies walk: trying to make their models feel more personable without letting them become pushovers.
Still, users may appreciate the quick response. OpenAI’s transparency about the issue, and its willingness to reverse course, suggests it’s taking user trust — and accuracy — seriously as the AI space continues to evolve.
