According to OpenAI, ChatGPT will now eliminate all dashes when directed to do so. These signs, which appear to signal text written by AI, have popped up everywhere in recent months, including school newspapers, emails, comments, customer service chats, LinkedIn posts, online forums, and ad copy. The inclusion of the em dash led people to criticize these writers for being lazy and relying on AI chatbots to do their work.
Of course, there are many people who argue for the em dash, saying it was part of the sentence long before LLM adopted the punctuation mark. But the fact that chatbots couldn’t seem to avoid using it has made the so-called “ChatGPT hyphen” a new and unpleasant addition to any text, even if it’s not a reliable signal of content created by generative AI.
This issue has been plaguing OpenAI for some time, as ChatGPT users have been unable to get the chatbot to stop using the symbol, even though they specifically requested the chatbot to stop using it.
Now, OpenAI CEO Sam Altman says the company has fixed the problem. “When you tell ChatGPT not to use em-dashes in custom instructions, it finally does what it’s supposed to do,” Altman said in a post on X, calling the update “a small but welcome victory.”
In a Threads post (which forced ChatGPT to apologize for “ruining the em dash”), the company explains that ChatGPT is better off not using em dashes if you instruct them not to use em dashes in custom instructions in your personalization settings. This means that em dashes will not necessarily be removed from the output by default, but at least you will have more control over how often they appear.
