ChatGPT Finally Stops Em Dash Overuse, Adds Group Chats

After years of users pleading for a simple change, ChatGPT has finally learned to curb its em dash habit. OpenAI CEO Sam Altman announced the update on X, calling it a “small-but-happy win” for writers and editors frustrated by the chatbot’s dramatic punctuation tendencies. Users can now instruct ChatGPT to avoid em dashes, and the AI will respect the setting in future responses.

The punctuation issue, once a minor quirk, became a widely recognized marker of AI-generated text. Even when users explicitly requested “no em dashes,” ChatGPT often ignored the instruction, prompting memes and online jokes. With this update, OpenAI has strengthened the chatbot’s “custom instructions” feature, giving users more control over tone, style, and now punctuation habits.

Beyond punctuation fixes, OpenAI has introduced Group Chats, a social feature that allows up to 20 users to interact with ChatGPT together. The AI can assist with brainstorming, answering questions, or mediating conversations, making the platform a collaborative space rather than just a solo tool. Currently available in Japan, South Korea, New Zealand, and Taiwan, the feature is in pilot mode and expected to expand globally.

This development is part of OpenAI’s broader push to enhance user experience and interactivity. Recent experimental launches, like Sora for short-form AI videos and Atlas, an AI browser for Mac, reflect the company’s goal to expand beyond simple chat and offer practical, creative tools.

With the em dash issue resolved and collaborative features underway, ChatGPT is evolving into a more versatile assistant, balancing precision, style, and social functionality—all while keeping users firmly in control of how it communicates.