ChatGPT is getting a mood slider — and with it, a small but telling shift in how we’ll think about conversational AI. Instead of folding tone and style into a long, fiddly prompt every time you want a different voice, OpenAI has added a simple set of personality controls that let you dial warmth and enthusiasm up or down, tweak how often the assistant uses emoji, and even lean on or away from structural flourishes like headers and bulleted lists. The company announced the rollout in mid-December, and the new options appear in the app’s Personalization settings.

For years, the art of coaxing a model into the “right” voice has been a secret handshake among power users: a mix of parenthetical stage directions, repeated instructions, and constant micro-editing. The new controls turn that craft into a set of app preferences. Choose “less warmth” once and the model will aim for emotionally neutral phrasing across conversations; choose “more enthusiasm,” and replies tilt brighter and crisper. There are three choices for each characteristic — more, less, or default — and OpenAI says the settings apply broadly across chats rather than only to the current thread.

You’ll find the switches in a place that feels intentionally ordinary: open the menu in the ChatGPT app, tap your profile, go to Personalization, and pick “Add Characteristics.” That deliberately consumer-facing placement matters. These aren’t hidden developer flags; they’re profile choices, the kind of small personalization people already expect in social or productivity apps. Making tone a persistent setting nudges the product away from being a one-off tool and toward being something that adapts to, and remembers, you.

The controls don’t replace the broader palette of built-in personalities OpenAI has been experimenting with — you can still pick a base style such as professional, quirky, candid, or friendly — but they let you fine-tune the emotional texture atop that base. Think of it as a wardrobe: a suit for the occasion, then a scarf or tie to change the mood. That combination makes it easier to mix and match a voice that fits the task — cautious and buttoned for a legal draft, or warm and emoji-friendly for a community update.

One of the practical nudges in this update is toward editing inside the chat. OpenAI has also added targeted editing tools so you can highlight a paragraph and ask for a constrained rewrite — “make this paragraph more concise,” “soften this sentence,” or “add a subject line” — without the old dance of referring to “the paragraph above.” That small change is what makes the warmth and enthusiasm switches actually useful: set the assistant to a low-enthusiasm, minimal-emoji mode and your email rewrites will follow that preference automatically.

On the surface, the changes look cosmetic — sliders for feelings and emoji counts — but they reflect a larger design judgment about how people want to live with assistants. Users are tired of treating prompt engineering as a lifestyle skill: it’s a brittle, repetitive way to get predictable behavior. Suppose the app can remember your stylistic defaults, the interaction model shifts from repeatedly instructing to quietly configuring. That matters because a consistent voice shapes trust and utility: an assistant that reliably hits your preferred tone can feel trustworthy in different contexts, from drafting a performance review to generating a playful social post.

There are trade-offs worth flagging. Making tone adjustable and persistent makes the assistant feel more personal, but it also raises expectations about continuity and control. How explicit should platforms be about when preferences are applied? Will a single “less warmth” toggle be too blunt for people who want nuance between professional and personal use? For now, OpenAI’s approach is conservative — clear toggles with simple options — but product teams will face pressure to add more contextual smarts (for example, switching tone automatically for emails versus brainstorming).

Designers and power users will likely welcome the change because it reduces friction: fewer repeated instructions, less manual cleanup, and a predictable baseline voice. Communication experts, however, may push back on the risk of flattening nuance; tone is often situational, and a persistent profile can be helpful or harmful depending on the task. For writers, the new features are an invitation to rethink workflows: set a default voice for rough drafts, then nudge tone for final edits rather than rewriting from scratch.

If you’re the sort who likes to experiment, the real fun will be in combinations. Pick a “professional” base, dial down enthusiasm, and remove emojis for terse, formal drafts. Or pick “quirky,” crank warmth up, and add emojis for idea-generation sessions. In either case, the important change is philosophical: tone is now an explicit product setting, not an emergent quirk of a single default voice. That shift makes it easier to stop treating conversational AI as a black box and start treating it like a personal tool you can configure once and forget — at least until you want something different.

OpenAI announced the controls publicly in mid-December and documentation about characteristics and personality customization is now live in their help center. Expect a steady trickle of press coverage and user-reaction threads as people test how well the sliders map to everyday writing problems. If you rely on ChatGPT for a lot of drafting or editing, it’s a small change that could shave minutes off repetitive work and remove one more place where prompt engineering gets in the way of productivity.

At its best, this feels like a modest but practical step toward making AI assistants feel less like temperamental tools and more like reliable collaborators. At its most consequential, it’s a reminder that how an AI speaks — not just what it says — is now part of the product people will set and expect. The new sliders don’t solve the deeper ethical and contextual questions around voice and authority, but they do make the emotional register of AI conversations an explicit choice. And for many users, that’s the point: less guesswork, more control, and one fewer awkward “please don’t use emojis in this reply” every morning.

Related /

• OpenAI launches GPT-5.2 as its new flagship AI model series
• GPT-Image-1.5 powers the new ChatGPT Images experience
• OpenAI and Anthropic are teaching AI chatbots to detect and limit underage users
• OpenAI debuts GPT-5.1 with expanded personality presets for all users