Since the launch of ChatGPT, users and editors have steadily picked up on certain stylistic quirks that made the model’s writing easily identifiable. One such quirk was the frequent and sometimes gratuitous use of the em dash. Writers noted that even when they explicitly instructed the tool not to use em dashes, the character kept slipping in. Acknowledging this, Sam Altman of OpenAI announced the update with a post saying, “Small-but-happy win: If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do!”
What the new setting actually does
According to OpenAI, the update adds a genuine writing-style preference: if the user specifies in the “Custom Instructions” section that ChatGPT should avoid em dashes, the model will now aim to comply. The company clarifies, however, that this doesn’t mean em dashes are globally removed by default – the instruction must be set. For many writers this represents more control over the chatbot’s output than before.
Mixed reviews from users and early testers
While the announcement was welcomed, reactions have been mixed. Some users reported immediate compliance: their subsequent responses from ChatGPT contained no em dashes. However, others posted screenshots showing the tool still inserting em dashes despite the instruction. One user described the experience as “a big but disappointing loss” after ChatGPT ignored the preference. At the same time, confusion remains over how to enable the setting and whether it applies consistently across all sessions.
Why this punctuation tweak matters more than it seems
At face value the em-dash update might look trivial, but it connects to deeper issues. For professional writers and editors, a tell-tale punctuation mark sniffed out by readers can undermine credibility. Allowing customization signals that AI writing tools must adapt not just functionally but stylistically to human norms. Moreover, the change poses questions around detectability: if style markers that flagged AI-written text become optional, authenticity checks become more complex.
What to keep an eye on going forward
Moving ahead the key questions are whether ChatGPT will comply consistently, and whether OpenAI will address other stylistic quirks beyond punctuation. Does this setting stay active across conversations and devices? Will the company add preferences for tone, syntax or vocabulary? For users and content-creators alike, watching how this feature rolls out and behaves is crucial.
Implications for writers and AI-generated text
The update signals a shift: user control is becoming as important as model capability. For those relying on AI for writing support, this means fewer unwanted quirks and more alignment with human style. But the change also reduces one of the easier “tells” of AI-authorship, complicating efforts to distinguish human vs machine writing. Ultimately, the “small but happy win” offered by OpenAI may mark the beginning of a broader era in which AI tools don’t just write, but write the way you tell them to.
Source: Live Mint
