OpenAI says it has resolved a longstanding issue that caused ChatGPT to continue using em dashes even when users explicitly instructed it not to. The punctuation mark, now widely referred to as the “ChatGPT hyphen” by critics, became one of the most recognizable signs of AI-generated writing as chatbots inserted it into school assignments, emails, customer support messages, marketing copy, LinkedIn posts and online discussions.
Although many writers have defended the em dash as a normal part of their style, its persistent and sometimes excessive appearance in AI outputs made it a point of frustration for users trying to produce text without it. For months, ChatGPT would ignore instructions to avoid the symbol, creating a mismatch between user preferences and the model’s behavior.
OpenAI CEO Sam Altman announced the fix in a post on X, saying that if users add “no em dashes” to their custom instructions, ChatGPT “finally does what it’s supposed to do.” Altman described the change as a “small-but-happy win.”
On Threads, OpenAI said ChatGPT will no longer default to using the punctuation when custom instructions specify otherwise. The company added a scripted apology from ChatGPT for “ruining the em dash,” highlighting how the update improves user control over stylistic preferences. OpenAI noted that the change does not remove em dashes by default, but users who rely on the personalization settings will now have a more consistent way to limit or eliminate them.
Featured image credits: Freepik
For more stories like it, click the +Follow button at the top of this page to follow us.