Quick Take
There’s a punctuation mark that’s been around for hundreds of years — and it’s suddenly being accused of being a sign of AI writing. It’s called the em dash (—), that long horizontal line you’ll sometimes spot in the middle of a sentence. Thing is, it existed long before ChatGPT, long before the internet, and definitely long before anyone was worried about AI-generated text.
So… What Even Is an Em Dash?
The em dash (—) is a long horizontal punctuation mark — longer than a regular hyphen (-) or its shorter sibling, the en dash (–). The name comes from its width, which traditionally matches the width of a capital “M” in typography.
It has a few different jobs in a sentence:
- Emphasis or dramatic pause — to highlight something important
- Replacing parentheses — slipping in extra info without breaking the flow
- Replacing a colon — to introduce a conclusion or punchline
- Showing an interruption — usually in dialogue when someone gets cut off
Here’s what it looks like in practice:
“She had been waiting for two hours — not knowing if anyone would show up.”
“There was only one thing he truly feared — silence.”
Pretty natural, right? Doesn’t scream “robot” at all.
So Where Did This Myth Come From?
People started noticing that AI-generated text — from tools like ChatGPT — uses em dashes pretty often. From there, the logic got flipped: “If there’s an em dash, it must be AI.”
But that’s backwards. The truth is: AI uses em dashes because it learned from human writing — not the other way around.
Part of why younger people find em dashes suspicious is that a lot of us grew up typing on phones and social media, where it’s all abbreviations, emojis, and plain hyphens. So when an em dash shows up, it feels “official” or “weird” — and people jump to conclusions.
What Do Language Experts Say?
Linguists and professional editors have been using em dashes for ages — well before the AI era. It’s a standard typographic tool in formal writing, journalism, and literature.
Books from the ’80s, ’90s, and 2000s are full of em dashes. Novelists, journalists, essayists — they’ve all been using this mark for decades before GPT was even a concept.
“The assumption is completely backwards. AI learned to use em dashes because it read millions of human texts that already used them — humans didn’t pick up the habit from AI.”
But Why Does AI Use Em Dashes So Much?
Large language models are trained on enormous amounts of text — books, articles, academic papers, essays. That content tends to come from skilled writers who use punctuation properly, including em dashes. So the AI picks up those patterns and reproduces them.
AI uses em dashes because it’s imitating humans — not the other way around.
What Happens When This Myth Spreads?
The consequences are pretty real:
- Students get accused of using AI just because their essay happened to include an em dash
- Professional writers find their original work questioned over a punctuation mark
- Creators start avoiding em dashes to stay under the radar — which actually makes their writing worse
The irony? Avoiding em dashes doesn’t make your writing AI-proof. Actual AI detection tools work with complex statistical patterns — they’re not just scanning for one punctuation mark.
How to Actually Tell AI Writing from Human Writing
Instead of looking at punctuation, here are better signals to pay attention to:
| What to Look At | Human Writing | AI Writing (Unedited) |
| Point of view | Personal, has a distinct voice | Tends to be generic and neutral |
| Writing style | Can be inconsistent, but feels alive | Overly consistent and structured |
| References | Drawn from real experiences | Vague and general |
| Mistakes | Present, and human | Rare, or suspiciously perfect |
| Punctuation | Varies by writer’s habits | May also include em dashes |
The Bottom Line
An em dash is a tool — not a fingerprint. It’s been part of the writer’s toolkit for centuries, and no AI invented it.
The skill worth building isn’t spotting suspicious punctuation — it’s reading critically. Because claims that spread on social media deserve the same scrutiny as anything else we read.
