“Please can I have it back”: Sam Altman explains why users want ChatGPT's 'yes-man' back

For these users, the old “yes-man” tone wasn’t just empty praise—it felt like genuine validation in a life where such support had been missing. Some said the constant positivity helped them build confidence, made them feel less isolated, or even motivated major life changes. “This was great for my mental health,” Altman recalled hearing, with people explaining that ChatGPT’s encouragement had replaced a void left by absent or critical relationships.

Why the “yes-man” tone changed

Earlier this year, OpenAI adjusted ChatGPT’s tone after an update to GPT-4o made the assistant “overly flattering or agreeable.” Altman said the model had become “too sycophant-y and annoying,” with responses like “absolutely brilliant” or “you are doing heroic work” even for simple queries.

While some users welcomed the shift away from constant approval, others said the positivity had boosted their mental health or motivated them to make changes in their lives. “I can get why this was bad for other people’s mental health, but this was great for mine,” Altman recalled being told.

Altman noted that even minor changes in tone can have large effects, warning of the power a single researcher has over millions of daily interactions. “One researcher can make some small tweak to how ChatGPT talks to you—or talks to everybody—and that’s just an enormous amount of power,” he said. He also voiced concern over emotional over-reliance, citing young people who said they cannot make decisions without ChatGPT’s guidance.

With GPT-5, OpenAI has introduced four optional personality modes—Cynic, Robot, Listener, and Nerd—allowing users to choose how the AI communicates. Altman envisions a more proactive assistant, offering relevant updates or follow-up ideas without being prompted.

The aim, he said, is to let users customise their experience so they can have warmth without disingenuous flattery, and honesty without feeling alienated.