OpenAI tunes GPT-5 to sound warmer and less robotic
OpenAI just tweaked GPT-5's personality to make conversations feel warmer and more approachable after feedback that it felt too stiff. The update introduces subtle touches like "Good question" without increasing sycophancy, according to internal testing. The rollout takes up to a day to complete.
If GPT-5 has felt a bit too cold and corporate lately, OpenAI is finally addressing the complaint.
A little warmth without the flattery
OpenAI announced on X that it is rolling out an update to make GPT-5 warmer and friendlier. The change comes directly from user feedback that the model previously felt too formal in daily conversations.
According to OpenAI, the adjustments are subtle rather than dramatic, but ChatGPT should feel noticeably more approachable now.
Users will start spotting small, natural touches like “Good question” or “Great start.” The company specifically noted that these phrases are meant to be genuine interactions, not superficial flattery.
To address concerns about sycophancy, OpenAI said internal tests showed no increase in people-pleasing behavior compared to the previous GPT-5 personality. The update may take up to a day to roll out completely, with more updates promised soon.
Why it matters
Tuning an AI's personality is always a tricky balancing act. Make it too stiff, and interacting with it feels like filing a tax form. Make it too friendly, and it turns into an annoying yes-man that blindly agrees with bad user prompts.
OpenAI proactively pointing out that sycophancy did not rise shows they know power users despise cheap praise. If they managed to thread the needle, everyday work inside ChatGPT might feel a bit more comfortable without losing precision.
Sources (1)
- @openai(opens in a new tab)Originalx.com