The overuse of em dashes is one of the most famous peculiarities of ChatGPT that OpenAI has resolved. The punctuation mark had become an unwanted AI-generated signature, showing up in school assignments, blog posts, and especially on LinkedIn.
The hyphen used by critics to identify lazy writers who relied too heavily on the chatbot was nicknamed the ChatGPT hyphen, even though the symbol has existed as a legitimate writing tool long before AI.
Users were frustrated by the issue for months. ChatGPT continued to produce em dashes even when specifically instructed not to. OpenAI’s community forums were filled with complaints from writers who simply couldn’t get the bot to comply.
That is why CEO Sam Altman now says the fix is here. Users can include the “no em dashes” preference in their personalisation settings—and this time, the model will actually listen. Altman called it a “small but happy win.” The company even made ChatGPT issue a tongue-in-cheek apology on Threads for “ruining the em dash.”
Still, this update doesn’t automatically eliminate the punctuation mark from all outputs. Instead, it gives users more control over how often it appears in AI-generated writing, offering relief to people tired of this easily spotted giveaway.
Yet, as some experts point out, punctuation alone isn’t what reveals AI writing. Commenting on the update, Oladotun Oguntola, a data analyst, noted that “tools like originality.ai or Turnitin can still detect very accurately even the most paraphrased AI content. Besides the em dashes, other common patterns—like the overuse of multiple adjectives (‘strong, policy-oriented impact statement,’ ‘concise, professional delivery’) or filler words such as ‘overall,’ ‘underscores,’ and ‘nuances’—easily give out contents written by AI.”
As AI-generated text becomes increasingly common, even small customisation options help users produce writing that feels more personal—and less like something stamped with the unmistakable fingerprints of a machine.
I am passionate about crafting stories, vibing to good music (and making some too), debating Nigeria’s political future like it’s the World Cup, and finding the perfect quiet spot to work and unwind.
One reply on “Has ChatGPT Really Removed This Famous Punctuation Mark?”
[…] noted that the model understands when to speak and when to stay in the background. Users can tag ChatGPT whenever they want input. It can also react with emojis and respond to messages that reference a […]