A punctuation symbol is being hit with serious ChatGPT accusations – and people aren't happy

Some people are only just learning about the em dash, or so it seems, as written content faces a new wave of ChatGPT-generated accusations – all thanks to the versatile punctuation mark.

Tweets, blogs, TikTok clips, memes and LinkedIn posts have all been birthed off the back of the debate, with some seemingly unaware that the em dash was around way before ChatGPT and modern technology as a whole.

For the blissfully unaware and for those who have rebranded the punctuation mark as the 'ChatGPT hyphen,' the em dash is a dynamic symbol that often steps in for commas, parentheses, colons, and semicolons. It's considered more interruptive or attention-grabbing than other marks, making it a popular choice for writers looking to add emphasis or highlight key information. Resembling a long horizontal line (—), it's longer than both a hyphen (-) and an en dash (–).

The first sighting reportedly dates back to the 1830s in the writing of C.S. Van Winkle and has since become a divisive piece of punctuation among writers. Essentially, the Marmite of the literary world – some love it, some hate it. Now, though, people are shying away from using it, worried that it'll make their work look like it was churned out by a bot.

It's ultimately become the gold or blue dress debate of 2025, and there are strong opinions all around.

"A shortcut for detecting if something is written with AI is they all use this symbol '—' throughout the writing. It's relatively rare when a human uses it, maybe once or twice, if that. But AI chats love using it. No clue why," one person claimed.

Another suggested: "The easiest way to spot if someone is using chatgpt to tweet, comment, or write anything online – the overuse of the em dash (—)."

Meanwhile, a third quipped: "anyone who uses an emdash is obviously ai—literally the most obvious tell."

But now, people are fighting back, rallying for justice for the innocent symbol caught in the heart of this controversy.

"No, the em dash is NOT a sign of ChatGPT — you just haven't been writing long enough to recognize the trend," Matthew Proctor, CEO of Narrative Bent, hit back. "To all of the writers and content creators in my network claiming the em dash is a dead giveaway of AI-generated content: please stop."

He continued: "The fact is, people have been using the em dash long before generative AI started adopting it. How do you think the algorithms 'learned' how to use them in the first place?"

Instagram

So, why does ChatGPT love the em dash so much?

Well, according to James Bore, chartered security professional and author of The Cyber Circuit, it's pretty much self-explanatory. The system is built by humans, and so, it picks up the language used during development.

James explains that ChatGPT tends to go for a more formal writing style, mostly because "that's a choice by the creators," and that particular style tends to be "more strongly governed by rules which make it easier for an LLM to produce convincingly".

Meanwhile, informal writing is more challenging to emulate realistically – of course, this includes using correct punctuation.

"The em-dash is not on most keyboards, and most people don't know how to type it (hold down the Alt key and type 0151, then release Alt), so its usage stands out," James tells Indy100. "It’s not that it uses the em-dash particularly often, given the writing style, just more than humans generally do."

Aside from generalised assumptions based on certain words and punctuation usage, James says there's no solid evidence or bulletproof software as of yet that a piece of content is 100 per cent AI-generated. However, there are rookie mistakes.

"A lot depends on the person using the LLM. I've seen people do things like leave in the prompts or the responses. Signs like ending long-winded posts with 'Did you have anything else I could help you with?' on social media are another hint," he explains.

"Most of the AI detection tools out there aren’t reliable, but there are suggestions that AI-written content has a lower entropy than human-written content," James continues. "This is going into information theory, but the entropy of any form of information is something there are ways to measure. More research is needed to really find out about this one, though."

Despite the challenges, ChatGPT's love for the em dash has actually had a positive knock-on effect on language learning.

"AI is definitely making people more conscious of how they write, not just to avoid being mistaken for AI-generated content, but also to preserve their unique voice and style," Anna Pyshna at Preply shares.

"This shift is encouraging writers to explore grammar rules, punctuation, and stylistic choices more thoughtfully. As a result, there's a renewed interest in the English language, with more people eager to understand its nuances and use it more authentically."

Long live the em dash!

You may also like...

• ChatGPT gets stressed at negative news just like us
• Should we really be using AI tools for therapy? Doctors reveal all

How to join the indy100's free WhatsApp channel

Sign up for our free Indy100 weekly newsletter

Have your say in our news democracy. Click the upvote icon at the top of the page to help raise this article through the indy100 rankings.