The Em—dash conspiracy: The AI punctuation wars that have sent LinkedIn into meltdown

Our latest YouTube video is all about the em—dash. And we’ve deliberately written this newsletter with loads of them.

A piece of punctuation that has basically ruined social media for all involved.

Once a modest punctuation mark—elegant, practical, and barely noticed—the em—new controversy: that it's a dead giveaway your content was written by AI. Em—dash is now at the centre of a classic social media war.

According to a growing crowd of digital sleuths, the em—dash has become the literary equivalent of a watermark—a supposed sign that ChatGPT had a hand in your blog post, CV, or LinkedIn thought-leadership thread.

Let’s pause—just like an em dash would ask us to—and unpack what’s going on here.

The em—dash is a piece of punctuation with range. It's named for being roughly as wide as the letter "M" and is distinct from the hyphen and en dash.

Writers love it because it’s expressive. It inserts a pause, breaks up rhythm, adds emphasis. It’s the closest thing to breathing on the page. You can use it instead of commas, colons, or parentheses—when you want to guide the reader's attention or change gears mid-sentence.

Different style guides argue over whether there should be spaces on either side (Chicago says no, AP says yes), but everyone agrees: it’s meant to support clarity and flow, not clutter it.

ChatGPT and other Large Language Models are trained on oceans of human-written content—blogs, essays, fiction, journalism. And that means they pick up human writing habits.

When asked to write in a casual, readable tone, AI often leans on em dashes. Not because it's obsessed with them, but because we are. Em—dashes are everywhere—from modern novels to marketing copy—and AI reflects those preferences back to us.

Some people claim AI overuses them, like seasoning from a robot chef—precise but soulless. Others argue it's just mirroring human writing. Either way, the em—dash isn’t a machine tell—it’s a stylistic trend we taught the machines to mimic.

The em—dash isn’t some new AI-age invention. It dates back to 16th-century printing, gained popularity in the 18th century, and became a favorite of literary giants like Emily Dickinson. (She was so into them, her poetry practically marched to a dash-dash beat.)

In the typewriter era, the em dash lost ground—you couldn’t easily type one, so people used two hyphens instead (--). But by the late 20th century, it made a comeback as writing grew more conversational and dynamic.

Fast forward to today: it’s everywhere again—newsletters, Substacks, screenplays, social captions. And yes, AI outputs.

One viral theory suggests the em—dash is a secret fingerprint of AI-generated text. Some even call it the “ChatGPT hyphen”.

But AI doesn’t use em—dashes because it’s trying to leave clues. It uses them because it thinks that’s what fluent human writing looks like. Especially in casual or persuasive formats. If you ask it to write a scientific abstract? The em dashes disappear. Instead, you get short, clear sentences and conservative punctuation.

AI adapts its style to match the tone and genre. More formal? Fewer dashes. More conversational? More dashes. The em—dash doesn’t expose the machine—it reflects the brief.

Despite all that, the myth persists. On social media, particularly TikTok and LinkedIn, people are treating em dashes like digital red flags. Spot a dash? Must be ChatGPT.

Brands have been accused of using AI because they dared to use one. Students worry their essays will be flagged.

It's all become a bit of a meme—part generational gap, part keyboard illiteracy, part dopamine-driven detective work. People are pattern-hunting. And once you believe something is a sign, you’ll start seeing it everywhere.

Let’s be honest: this em dash panic has nothing to do with punctuation and everything to do with anxiety.

In a world increasingly filled with synthetic content, people are desperate for clues. Any sign—no matter how minor—that separates human from machine feels like a lifeline. And punctuation? It’s easy to latch onto. It feels scientific. But it’s not.

Some of this stems from confusion around “AI watermarking.” Yes, researchers have explored ways to statistically fingerprint AI text—but no, that doesn’t mean ChatGPT is deliberately planting em dashes. That theory has zero basis in fact.

Short answer: no, the em dash is not a reliable AI detector.

Longer answer: relying on a single punctuation mark to detect AI is like using sunglasses to identify spies. It might be a clue—but it’s probably coincidence.

Even OpenAI scrapped its AI detection tool due to accuracy issues. Linguists and editors alike say the same thing: humans use em dashes. A lot.

In fact, the people most eager to police writing for signs of AI—those rigid about style, grammar, and tone—often sound more robotic than the AI they’re trying to expose.

Here’s what the em dash discourse actually tells us: we’re scared. Not of punctuation—but of losing the ability to tell what’s real. We want clean lines, simple answers, obvious signals.

But writing doesn’t work that way. And neither does authorship.

So if you see an em—dash, don’t panic. Don’t assume it came from an algorithm. Maybe it did. Maybe it didn’t. Either way, one dash doesn’t tell you who wrote something—just how they wanted it to feel.

And if it feels human—that might be the only test that really matters.

Does this email? There might be one too many —.

The em—dash has lived many lives: poetic, practical, old-fashioned, revived. Now, it’s become a symbol in the fight over authenticity. But it’s not the true villain.

It’s just punctuation. Stylish, flexible, and—despite the conspiracy theories—completely innocent.

So the next time someone posts about the “AI dash,” take a moment. Maybe even a dash-shaped pause. And then carry on writing—em dashes and all.