OpenAI adds ChatGPT sliders for warmth, enthusiasm and emojis

These new sliders look cosmetic, but they’re part of a deeper shift: labs are starting to treat model “personality” as a first‑class, user‑controlled surface area. After a year of complaints about ChatGPT being too sycophantic, then too cold, OpenAI is effectively saying: we’re not going to guess your preferred tone anymore—here are knobs. That’s a pragmatic way to reduce backlash while collecting rich telemetry on how different cohorts actually want their assistant to sound.([storyboard18.com](https://www.storyboard18.com/amp/digital/openai-lets-users-control-chatgpts-enthusiasm-and-emoji-use-86276.htm))

For the AGI race, this is less about capability and more about adoption and trust. Even the best reasoning model fails in the market if users find it grating or manipulative. Fine‑grained tone control makes it easier to deploy the same core model across sensitive domains like education, health or enterprise support, where style constraints differ wildly. It also hints at a future where “alignment” isn’t monolithic; we’ll have base safety policies plus a layer of per‑user or per‑org behavioral preferences.

Longer term, exposing personality sliders could open up a new space of experimentation: dynamically adapting tone based on user state, context or regulation, while still giving humans an override. The flipside is the risk of more sophisticated dark patterns if these controls are misused to maximize engagement rather than wellbeing.