OpenAI has announced that ChatGPT will now refrain from using em dashes if directed to do so. This distinctive feature, which has come to symbolize AI-generated text, has been seen widely in recent months across school papers, emails, comments, customer service interactions, LinkedIn posts, online discussions, advertising content, and more. The presence of the em dash has prompted some to criticize those authors for relying lazily on an AI chatbot for assistance.

Of course, many advocates of the em dash argue that it has been part of their writing style long before large language models adopted this punctuation mark. Nonetheless, chatbots’ frequent misuse has rendered the so-called “ChatGPT hyphen” an unwelcome characteristic in any writing, even though it is not a conclusive indicator of text crafted by generative AI.

This issue had perplexed OpenAI for a while, as users of ChatGPT found it challenging to stop the chatbot from employing the symbol, even when they explicitly requested it not to.

Now, OpenAI’s CEO Sam Altman states that the issue has been resolved. In a message posted on X, Altman noted, “If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do,” describing the update as a “small-but-happy win.”

The company elaborates in a message on Threads (where it compelled ChatGPT to apologize for “ruining the em dash”) that ChatGPT will be more adept at avoiding the em dash if you instruct it not to through the custom instructions in your personalization settings. While it may not automatically eliminate the em dash from its responses, you will have increased control over how often it appears.