If you’ve ever used ChatGPT or read something written with its help, you’ve probably noticed a long dash popping up more than usual. That punctuation mark is called the em dash, and while it might look trendy or even suspicious to teachers, it has a real purpose.

Unlike a hyphen (-) or an en dash (–), the em dash doesn’t connect words or indicate ranges. Instead, it signals an interruption — a stylistic pause stronger than a comma but smoother than parentheses.

For students learning to write, it can make writing flow in a more natural and somewhat conversational way.

ChatGPT is a LLM, short for language learning model, meaning it is trained on massive amounts of text: books, articles, essays and conversations. Most of the time, these references are written by many professional writers who rely on em dashes to make sentences clearer.

As a result, the model learned that the em dash is the best way to connect ideas while keeping writing readable — especially since it naturally occurs in the language. It helps avoid messy sentence structures and allows explanations to be read naturally.

However, the em dash has become something of a red flag in classrooms. Because ChatGPT uses it often, some teachers immediately assume AI-generated writing — even though plenty of human writers use it too.

“I love the em dash, I think it’s a great piece of punctuation,” said Naomi Kirchner, English teacher. “It’s curious, I have seen more em dashes since students started using AI.”

ChatGPT’s frequent use of the em dash reflects how language online is evolving. Artificial intelligence systems are trained on huge collections of human writing — novels, blogs, news articles — where this punctuation has grown in popularity.

Many students are worried about using the em dash due to possible repercussions. 

“I am worried because AI detectors flag the em dash, and I would be falsely accused of using AI,” said sophomore Grayson Kerr.

Interestingly, some AI detectors now look at features like em dash frequency to judge whether a piece is written by a human or an AI. Because ChatGPT tends to favor the em dash, a student using it often might accidentally trip a detector’s “AI-style” alert — even when the work is completely original, showing how difficult it can be to define authentic writing when human and AI styles are starting to overlap.

As an example, this article was handwritten, yet AI detectors still say that it was over 20% AI written.