On November 14, Sam Altman, the CEO of OpenAI, made a seemingly simple post on social media. He celebrated a "small-but-happy win," noting that if users instructed ChatGPT not to use em-dashes in custom instructions, the AI would finally comply. However, this minor update triggered a massive online discussion, uncovering a widespread phenomenon: punctuation anxiety in the age of AI.
The Viral Debate Over a Tiny Dash
Altman's announcement was quickly met with a wave of responses from users across platforms like X and Reddit. Many shared screenshots proving ChatGPT was still sneakily inserting the long dash (—) into its responses, calling it a "peak betrayal." Others joked about their own "little victories" in battling the AI's stylistic quirks.
The conversation revealed a deeper cultural shift. Users like Simon Hedlin expressed hope that people who genuinely enjoy using the "most versatile punctuation mark" could do so again without being accused of using AI. The core issue became clear: the em dash has become a tell-tale sign of AI-generated text, forcing human writers to reconsider their own style.
Indian Writers Deliberately 'Dumbing Down' Their Text
This new punctuation anxiety is acutely felt among young professionals and students in India. Rohan Jha, a 29-year-old from Delhi working in communications, often finds his family questioning his use of sophisticated punctuation and phrases in casual texts. "I usually avoid punctuation like em dashes when chatting with family or friends. In daily life, I prefer simple, natural language," he explains.
The fear of sounding like a bot is real. Riddhi Mukadam from Noida agrees, stating, "I avoid using em dashes now because they look very 'ChatGPT-ish'. Too many, and it starts to feel robotic." Similarly, 21-year-old Prachi Bisht admits to skipping such marks not because they are wrong, but because they sound "too polished."
Bhumika Rawal (22) extends this caution to other symbols. "I sometimes deliberately avoid using the tilde symbol, semicolon, or an em dash because people label it as AI-generated content. Even quoting text to emphasise a point now feels like an AI style," she says.
How AI Learns and Why It Loves the Em Dash
An AI expert explains that models like ChatGPT and Gemini are not directly taught grammar rules. Twinkle Chawla, an AI data scientist, clarifies, "Through predicting the next word, the model picks up where punctuation naturally appears — like a full stop after a thought or an em dash for a break." The AI learns statistically from billions of words from books and websites, adopting patterns found in refined writing.
"AI models don't consciously choose punctuation styles — they mirror patterns in the text they're trained on," Chawla adds. Since em dashes frequently appear in essays and literary non-fiction, the model uses them in formal or expressive contexts. This pattern has become so pronounced that a Washington Post analysis of over 328,000 messages from GPT-4o found em dashes appearing in more than half of the messages by mid-2025.
Other markers of AI text identified in the study include frequent use of emojis (like the green check mark and brain), a conversational tone with contractions, and the repeated phrase pattern "not just X, but Y."
The Human Response: Embracing Imperfection
In response, people are actively altering their writing to sound more human. This includes dashing dashes for commas, switching to chaotic lowercase, overusing exclamation marks, and intentionally breaking grammar rules.
"I try not to make my sentences too perfect," says Riddhi. "I skip fancy punctuation, break sentences more, and sometimes add a 'hmm' or 'tbh' just so it sounds more like how I'd actually talk. If something reads too clean, it instantly feels like a bot wrote it."
Mukadam reflects on the irony: "Perfect grammar used to be a good thing, but now it feels a bit emotionless. When everything's polished, it loses that raw, human touch. People make typos, skip commas, use emojis — that's what makes writing feel alive."
The internet is divided. Some, like one X user, have stopped using em dashes altogether to avoid suspicion. Others point out that the link between AI and the dash exists only because many people today are unfamiliar with advanced punctuation. Meanwhile, a segment of users admits they only discovered the em dash through ChatGPT's prolific use of it.
Ultimately, Sam Altman's post about a small technical fix opened a large window into how AI is reshaping not just what we create, but how we communicate our very humanity. The humble em dash has become a symbol of the struggle to remain authentic in a world of increasingly perfect, and perfectly suspicious, prose.
