You just finished a piece of writing that felt effortless. The sentences flowed, the rhythm was perfect, and you dropped an em dash exactly where it belonged—a punctuation mark you learned to use with pride. Then the comment lands: “This reads like AI.”
Your stomach drops. You weren’t using a bot. You were just writing well. But in 2025, writing well is indistinguishable from writing artificially. And that paradox is slowly strangling the joy out of expression.
The modern writer’s nightmare isn’t the blank page—it’s being accused of not being human.
Consider the story of a non-native English speaker who learned the em dash as a mark of advanced writing. “Actually, I was taught to use em-dashes for advanced writing when I was learning English as a foreign language. But these days, people say it looks like AI and tell me not to use them. It’s really confusing.”
That confusion is the new normal. The same tool that once signaled mastery now signals automation. And the solution being offered? Write worse. Avoid the em dash. Break up your sentences. Add a few passive clauses. Sound more like a machine so that other machines don’t think you’re a machine.
We are collectively sacrificing linguistic richness on the altar of human verification.
This isn’t about grammar. It’s a cultural immune response to algorithmic homogenization. The em dash has become a biomarker—a telltale sign that your prose was generated by a model trained on every writer who ever used one. But here’s the irony: those models learned the em dash from us. We wrote the training data. Now we’re being punished for our own signature.
Take a side: this is a tragedy. The war on em dashes is a mistake—a symptom of our broken relationship with AI. We are so terrified of being flagged as inauthentic that we’re willing to amputate our own style. We’re dumbing down our writing to prove we’re not stupid.
And it’s not just the em dash. It’s the well-placed semicolon. The rhythmic repetition. The parallel structure. Every flourish that made writing human is now suspect. Writers are editing their own voice out of fear of detection. They’re becoming their own AI detectors.
The most dangerous sentence in modern writing is the one that’s too good to be true.
So what do you do? Do you self-censor? Do you adopt a flat, monotone style to pass the Turing test? Or do you write with your full voice and risk being flagged as a bot?
Choose the latter. Write the em dash. Use the long sentence. Let your prose breathe. Because the moment we start writing for machines—even to prove we aren’t one—we’ve already lost. The algorithm doesn’t need us to be readable. It needs us to be detectable. And the only way to stay undetectable is to be artless.
Be artful. Be human. Let them accuse you. The accusation is a compliment in disguise.
FAQ
Q: Is this really a big deal? Aren't a few people overreacting to the em dash thing?
A: It's not just the em dash. It's a broader pattern where writers are being forced to suppress their natural voice to avoid being flagged as AI. This affects anyone who writes online—journalists, bloggers, students, even authors. The anxiety is real, and it's reshaping how we communicate.
Q: So should I stop using em dashes to avoid being accused of using AI?
A: No—that's exactly the wrong move. If you stop using the em dash, you're letting the algorithm dictate your style. Write the way you naturally write. The fear of being flagged is a trap; the only way to fight it is to write with boldness, not self-censorship.
Q: Maybe the AI detectors are right? Couldn't it be that em dashes are actually a sign of lazy AI writing?
A: Consider the source: AI detectors are notoriously unreliable and often penalize human writers who write well. The em dash is a legitimate punctuation mark with a long history. If you're using it skillfully, it's not lazy—it's elegant. The problem isn't the tool; it's the paranoia that surrounds it.