In November 2025, OpenAI made an announcement stating that ChatGPT had successfully tackled the issue of excessive em dash usage in its text - generation process. In the past, numerous users had reported that ChatGPT frequently disregarded instructions to "steer clear of using em dashes." This led to a situation where writing styles became overly uniform and even became the subject of jokes on social media platforms.

Following this update, users now have the ability to disable em dashes by utilizing custom instructions. The system is designed to precisely adhere to these requirements, thereby enhancing the personalized user experience. OpenAI's CEO, Sam Altman, referred to this as a "minor yet delightful enhancement." He stressed that this optimization was made possible by improving the custom instruction features, which in turn gives users more control over how punctuation marks are used.