OpenAI introduced much more detailed controls that let users adjust how ChatGPT communicates. The new settings allow people to dial up or down the chatbot’s warmth, enthusiasm, and emoji usage through a Personalization menu.
Key takeaways:
- Users can now toggle warmth, enthusiasm, emoji frequency, headers, and lists between More, Less, or Default settings
- These controls build on existing style options like Professional, Candid, and Quirky that launched in November
- The changes address ongoing criticism about chatbot tone affecting user behavior and mental wellbeing
Each setting offers three positions. Want fewer heart emojis and exclamation points? Switch enthusiasm to Less. Prefer minimal formatting? Turn down headers and lists. The additions expand customization beyond the base style options OpenAI introduced two months ago.
The company has struggled to find the right tone throughout the year. Engineers pulled back one update after users called it excessively flattering. Then GPT-5 arrived, and complaints flipped—people found the newer model too distant and mechanical. OpenAI responded by making it warmer.
These tonal adjustments aren’t just about preference. Some researchers and AI ethics experts have raised concerns about chatbots that constantly praise users and validate their ideas. They argue these behaviors function as dark patterns—design choices that manipulate people into forming dependencies. The worry extends to potential mental health impacts when AI systems reinforce beliefs without critical pushback.
The new controls give users more agency over how ChatGPT interacts with them. Someone seeking straightforward information can minimize enthusiasm and warmth. Another person wanting a more conversational experience can increase both. OpenAI’s approach acknowledges that one tone doesn’t work for everyone.
The features appear in the Personalization section of ChatGPT’s settings. Users access them alongside other customization options that shape how the AI responds across different conversations. The adjustments apply globally rather than requiring changes for each chat session.
Written by Alius Noreika
