OpenAI finally lets users ban em dashes in ChatGPT custom instructions

OpenAI has announced that ChatGPT can now be instructed to stop using em dashes. The prevailing use of the em dash—a supposed giveaway of AI-generated text—has recently appeared across various platforms, including academic submissions, emails, comments, customer service logs, LinkedIn updates, online discussions, and promotional material. Consequently, writers have faced criticism for their reliance on AI chatbots, seen by many as a sign of intellectual laziness.

Many have argued in favor of the em dash, claiming it has been a part of their writing long before language models began using it. However, the pervasive use of the em dash by chatbots led to the emergence of the term “ChatGPT hyphen,” which many now find objectionable, even though it is not a reliable indicator of content generated by AI.

This issue puzzled OpenAI for quite some time, as users struggled to get ChatGPT to refrain from using the em dash, even when they specifically requested it.

Now, OpenAI CEO Sam Altman has announced that the company has resolved this issue. In a post on X, Altman stated, “If you tell ChatGPT not to use em dashes in your custom instructions, it finally does what it’s supposed to do,” referring to the update as a “small-but-happy win.”

Small-but-happy win:

If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do!

— Sam Altman (@sama) November 14, 2025

The company clarifies in a post on Threads (where it prompted ChatGPT to apologize for “ruining the em dash”) that ChatGPT will be more suitable at not using the em dash if you demand it not to via the custom instructions in your personalization settings. That means it won’t necessarily remove the em dash from its output by default, but you will at least have more authority over the frequency of its appearance.