OpenAI is turning one of ChatGPT’s most subjective qualities into something you can actually dial in. Instead of living with a single default voice, you can now decide how warm, enthusiastic, and expressive the assistant should sound, and even rein in its emoji habit if you prefer a more buttoned‑up style. It is a small interface change with big implications for how people relate to AI, shifting tone from a fixed design choice into a user setting.
At a practical level, this means the same underlying model can present itself very differently depending on what you are doing, from a reserved research aide to a bubbly brainstorming partner. It also signals a broader shift in AI design, where personality is treated less as a marketing gimmick and more as a configurable layer that sits on top of core capabilities. I see this as the start of a new phase in consumer AI, where users expect to shape not just what the model does, but how it sounds while doing it.
What exactly changed in ChatGPT’s tone controls
The latest update adds a dedicated set of sliders and toggles that let you tune how expressive ChatGPT should be, instead of relying on one-size-fits-all defaults. Inside the app’s Personalization area, there is now a Characteristics section that exposes controls for warmth, enthusiasm, and how emotionally colored the assistant’s replies feel, so you can move from clipped and neutral to expansive and upbeat without touching the underlying model. That Characteristics panel is described as a way to decide how “expressive the assistant should be,” which turns what used to be a hidden system prompt into a visible user choice backed by the new granular tone controls.
On top of that, OpenAI is giving people simple presets for personality intensity, with options to make responses more, less, or default in terms of emotional color. Instead of manually prompting “answer in a formal tone” every time, you can set a baseline personality and let the system carry it across chats, which is especially useful if you bounce between the web interface and the official mobile app. Reporting notes that these personality and tone controls are available across the main ChatGPT experience and the official mobile app, with the update posted by Shalom Levytam as a significant expansion of user control.
Where to find the new warmth and enthusiasm settings
OpenAI has tucked the new tone options into a place that already made sense for casual users, rather than burying them in developer-style configuration menus. To reach them, you go into the Personalization tab in your ChatGPT settings, then open the Characteristics section, where you can adjust how warm or reserved you want the assistant to sound. That flow is described as a straightforward way to “navigate” into the new Enthusiasm Adjustment Features, with the controls framed as part of a broader push to make the assistant feel more tailored without requiring technical knowledge, which is why the Enthusiasm Adjustment Features live alongside other Personalization settings.
Once you are in that Characteristics panel, the interface presents simple choices rather than a wall of sliders, so you can pick “more,” “less,” or “default” for enthusiasm and warmth without needing to interpret percentages. Coverage notes that OpenAI is explicitly letting users decide between those three options to adjust ChatGPT’s personality, and that these switches are found in the Personalization tab that now acts as the central hub for tone, emoji, and style preferences. Jackson Chen highlights that this is meant to be a user-facing control, not a developer-only feature, which is why the “more, less or default” personality options are framed as everyday choices rather than advanced configuration.
How the enthusiasm dial actually works
Under the hood, the new enthusiasm control is essentially a high-level instruction that steers how effusive the model is allowed to be in its wording and pacing. When you set enthusiasm to “more,” you are telling ChatGPT to lean into positive language, exclamation points, and supportive phrasing, while “less” nudges it toward shorter, more matter-of-fact replies that strip out verbal cheerleading. Reporting describes this as a direct way to adjust ChatGPT’s enthusiasm level, with the system treating it as a persistent preference that shapes responses across topics, which is why the update is framed as letting users directly adjust the enthusiasm level rather than relying on ad hoc prompts.
In practice, that means the same question can feel very different depending on your setting: a “more” profile might respond to a career win with extended congratulations and suggestions for next steps, while a “less” profile might simply acknowledge the fact and move on to the requested analysis. The system also ties enthusiasm to warmth and emoji use, so dialing things down tends to reduce not just exclamation marks but also the emotional framing around advice, which some users find more trustworthy in professional contexts. OpenAI has said that users can tweak warmth, enthusiasm, and emoji use together, with one report summarizing that ChatGPT users can now adjust the chatbot’s warmth, enthusiasm, and emoji use in a single place, based on a social media post from Ope describing the new controls.
Why warmth and emojis suddenly matter
For years, people have complained that AI assistants either sound too robotic or too chirpy, with little room in between, and OpenAI is clearly responding to that feedback. Some users felt the model was distant and clinical, while others were put off by overly friendly phrasing that did not match the seriousness of their questions, especially in areas like health or finance. The new controls are explicitly framed as a response to those complaints, with one report noting that OpenAI is letting users adjust ChatGPT’s tone, warmth, and emoji use after hearing that the model felt distant, and that this is meant to bridge the gap between sterile and overly familiar by giving people a way to adjust tone and warmth themselves.
Emojis are a surprisingly big part of that equation, because they act as a shorthand for emotional intent in chat interfaces, and not everyone wants them in the same way. Some people like a sprinkle of emojis in casual conversations or creative brainstorming, while others see them as unprofessional in work settings or distracting in technical explanations. The update acknowledges that split by letting users not only reduce emoji use but also exclude emojis entirely if they want a text-only assistant, with coverage pointing out that the new customization options include the ability to control warmth, enthusiasm, and emojis, and even an option to exclude emojis entirely for those who prefer a more formal style, as described in a piece on how users can customize a warmer or more enthusiastic bot.
From one-size-fits-all to granular personality
What makes this update more than a cosmetic tweak is the shift from a single default persona to a spectrum of personalities that users can mix and match. Earlier iterations of ChatGPT already experimented with named personas like GPT Cynic, Listener, and Nerd, as well as Professional and Quirky options, which gave people a taste of how different tones could change the feel of the same underlying intelligence. Those personalities did not always make the model feel more human, but they did show that tone is a powerful lever, and the new controls build on that by letting you adjust warmth and enthusiasm on top of whichever persona you pick, as noted in an opinion piece that describes how GPT personalities like Cynic, Listener, and Nerd affect tone.
I see this as part of a broader trend in AI where personality is treated as a modular layer rather than a fixed brand identity. Instead of saying “this is what ChatGPT sounds like,” OpenAI is effectively saying “this is what ChatGPT can sound like, and you choose the version that fits your context.” That is a subtle but important shift, because it acknowledges that a student using GPT Nerd for homework might want a different emotional register than a manager using a Professional persona to draft performance reviews, and it gives both of them tools to fine-tune warmth and enthusiasm without switching models. The result is a more granular personality stack, where named personas like Cynic or Listener sit alongside user-controlled sliders for tone and emoji use, creating a layered approach to how the assistant presents itself.
Everyday scenarios where tone control changes the experience
The impact of these controls becomes clearer when you look at specific use cases, from classrooms to customer support. A teacher using ChatGPT to generate quiz questions might prefer a neutral, low-enthusiasm tone that keeps the focus on content, while a student using it as a study buddy could benefit from a warmer, more encouraging style that celebrates small wins and nudges them to keep going. In a corporate setting, a project manager drafting stakeholder emails may want a professional, low-emoji voice, while the same person planning a team offsite could switch to a more upbeat, emoji-friendly profile to match the informal nature of the event, which is exactly the kind of flexibility the new warmth and enthusiasm controls are meant to unlock as part of the broader ability to control warmth, enthusiasm, and emojis.
Customer-facing roles are another obvious beneficiary, because tone can make or break how a response is perceived even when the content is correct. A support chatbot that is too cheerful when delivering bad news can feel tone-deaf, while one that is too cold when resolving a problem can come across as uncaring, and both extremes can erode trust. With the new settings, a company could configure its internal ChatGPT instance to use a measured, empathetic tone with limited emojis for support interactions, while allowing marketing teams to crank up enthusiasm for campaign brainstorming, all without retraining the model. That kind of scenario highlights why giving users direct control over warmth and enthusiasm is not just a cosmetic feature but a practical tool for aligning AI behavior with human expectations in different contexts.
How this fits into OpenAI’s broader strategy
OpenAI’s decision to expose tone controls in the main ChatGPT interface fits a pattern of gradually handing more steering power to end users instead of keeping it behind API parameters. Earlier personalization features focused on things like preferred writing style or domain expertise, but warmth and enthusiasm touch on something more intimate, which is how the assistant feels to interact with over time. By putting these options in a prominent Personalization tab and labeling them clearly as Characteristics, OpenAI is signaling that tone is now a first-class part of the product, not an afterthought, and that it expects people to treat emotional style as a setting they revisit, much like notification preferences or theme choices, which is why the Characteristics section in Personalization is framed as a core feature.
There is also a strategic angle in how these controls are rolled out across platforms, including the official mobile app, which is where many casual users first encounter ChatGPT. By making personality tuning available on phones as well as the web, OpenAI is positioning ChatGPT as a personal companion that you can shape to your liking, rather than a static tool that only lives in a browser tab. That aligns with the company’s broader push toward more persistent, personalized AI experiences, where your preferences for warmth, enthusiasm, and emoji use follow you across devices and sessions, and where future updates could layer in even more nuanced controls for things like humor, formality, or cultural references, building on the current set of granular personality and tone controls that already span the main app surfaces.
The limits of personality tweaks and what comes next
Even with all these new knobs to turn, there are clear limits to what tone controls can achieve, and some early commentary has been blunt about that. Changing warmth or enthusiasm does not alter the model’s underlying reasoning or factual accuracy, and it does not magically make the assistant “feel” human in any deep sense, even if the surface-level chat becomes more pleasant. One analysis argues that the new personalities, including GPT Cynic, Listener, and Nerd, do not really make ChatGPT feel more human, because they mostly adjust phrasing and attitude rather than the substance of how the model thinks, which is a useful reminder that new personalities do not change the core model.
Looking ahead, I expect OpenAI and its competitors to keep layering more sophisticated controls on top of these basics, perhaps letting users define custom tone profiles for different contexts or share presets the way people share Lightroom filters or Discord themes. There is also room for more dynamic behavior, where the assistant automatically dials down enthusiasm when discussing sensitive topics or ramps it up when celebrating milestones, guided by user consent and clear settings. For now, though, the ability to tune how warm and enthusiastic ChatGPT sounds, and to decide whether emojis belong in your conversations at all, already marks a meaningful step toward AI that adapts to people rather than forcing people to adapt to it, which is exactly what the latest round of enthusiasm and warmth controls is designed to deliver.
More from MorningOverview