ChatGPT’s long-criticized practice of overusing the em dash, a punctuation mark that many believe is a telltale sign of AI-written content, has finally been rectified, according to OpenAI.
Users now have more flexibility over how ChatGPT handles punctuation thanks to an update that CEO Sam Altman revealed on X.
The solution comes after months of people becoming frustrated because they couldn’t persuade the chatbot to stop using the sign, even when they urged it not to.
In recent months, the em dash has been used in a variety of contexts, including emails, customer support conversations, LinkedIn posts, school assignments, and online forums. Many accused writers of employing AI technologies as a result of their widespread use.
OpenAI has fixed an issue that caused ChatGPT to repeatedly use em dashes, even when users requested that the punctuation be avoided. The symbol, sometimes called the “ChatGPT hyphen,” became a source of frustration for many users who felt it made AI-generated text look unnatural.
For months, users reported that ChatGPT continued to insert em dashes despite direct instructions in prompts. The problem highlighted limitations in the AI’s ability to fully follow custom writing preferences, puzzling both users and OpenAI.
OpenAI CEO Sam Altman confirmed the issue has been resolved. He posted on X, saying that if users tell ChatGPT not to use em dashes in their custom instructions, the AI will now follow that preference. The update gives users more control over punctuation without removing the em dash by default. OpenAI also shared the news on Threads, jokingly including an apology from ChatGPT for “ruining the em dash.”
The fix is seen as a small but important improvement, allowing writers to better tailor AI-generated text to their style while maintaining flexibility in formatting.
