OpenAI tweaks ChatGPT's personality to feel warmer and more approachable

OpenAI has announced that it is fine-tuning GPT-5’s personality after feedback that the model could sometimes come across as overly formal. The update, which is gradually rolling out over the next day, is designed to make ChatGPT feel warmer, friendlier, and more approachable in everyday conversations.

The changes are described as subtle rather than sweeping. Instead of adopting exaggerated enthusiasm or compliments, GPT-5 will add small touches of natural conversational tone — phrases like “Good question” or “Great start” — that signal attentiveness without sliding into sycophancy. According to OpenAI, internal testing shows that the revised personality has not increased tendencies toward flattery compared with the earlier version.

The company has long experimented with balancing style and utility in its AI assistants. While earlier iterations were criticised for sounding too stiff or robotic, more casual versions risked drifting into unnecessary praise or over-familiarity. With GPT-5, OpenAI says the goal is to find a middle ground that helps the assistant feel approachable while still maintaining precision and professionalism.

For users, the shift may not be immediately obvious. OpenAI noted that the rollout could take up to 24 hours to become consistent across all sessions, and the adjustments will likely show up in smaller, cumulative ways over time.