OpenAI says it has finally fixed ChatGPT’s long-criticised habit of overusing the em dash, a punctuation mark many considered a giveaway of AI-written text.
The update, announced by CEO Sam Altman on X, gives users more control over how ChatGPT handles punctuation.
The fix follows months of frustration in which users were unable to get the chatbot to stop using the symbol, even when explicitly told not to.
The em dash has appeared everywhere in recent months: school assignments, emails, customer service chats, LinkedIn posts and online forums. Its widespread use sparked criticism, with many accusing writers of relying on AI tools.
At the same time, others defended the punctuation mark, arguing it long predates AI and has always been part of their natural writing style.
But ChatGPT’s inability to avoid it, despite direct requests, made the so-called “ChatGPT hyphen” a new point of frustration.
Why em dash became problem
For months, ChatGPT users tried to force the chatbot to stop using the em dash.
Despite repeated instructions, the model continued inserting the symbol, making it seem like a signature of AI-generated text.
The issue puzzled OpenAI, as even strict prompts failed to suppress the punctuation.
OpenAI says issue is now fixed
OpenAI CEO Sam Altman said the company has resolved the problem. In a post on X, he wrote: “If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do,” calling it a “small-but-happy win.”
Small-but-happy win:
— Sam Altman (@sama) November 14, 2025
If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it's supposed to do!
On Threads, OpenAI explained that ChatGPT will be better at avoiding the em dash if users set the preference in custom instructions. The company even jokingly forced ChatGPT to apologize for “ruining the em dash” as part of the announcement.
However, the update does not remove the punctuation by default. It simply gives users more control over how often ChatGPT uses it.
