ChatGPT Finally Stops Overusing Em Dashes as OpenAI Rolls Out Fix
OpenAI has addressed a longtime frustration among users by finally resolving ChatGPT’s widely criticized habit of overusing the em dash. The punctuation mark became so closely associated with AI-generated writing that many users began treating it as a telltale sign of content produced by ChatGPT.
The update was confirmed by OpenAI CEO Sam Altman, who announced on X that the issue had been fixed. He noted that the model will now follow user preferences more reliably when asked to avoid specific punctuation. For months, people repeatedly instructed the chatbot not to use the em dash, but it continued inserting the symbol even during strict writing guidelines.
The problem grew so noticeable that screenshots and complaints filled social platforms, with some users describing the em dash as ChatGPT’s “signature move.” It showed up in school assignments, work emails, customer service scripts, and social media posts. The consistency of its usage led many to assume that anyone overusing the punctuation was likely relying on AI.
At the same time, several writers defended the em dash, pointing out that it has been part of natural writing styles long before the rise of AI. But ChatGPT’s difficulty in avoiding it, despite direct requests, turned the punctuation into an unintended controversy.
OpenAI acknowledged that the inconsistency puzzled its engineers because clear instructions should have prevented the model from using the symbol. According to the company, the latest update gives users more precise control through custom instructions. Altman called the improvement a “small-but-happy win,” signaling that the larger issue was rooted in how the model handled stylistic constraints.
On Threads, OpenAI added a humorous twist by having ChatGPT apologize for “ruining the em dash,” while also clarifying that the punctuation is not being removed from its writing capabilities. Instead, the model now respects user preferences more accurately, allowing those who want to avoid the symbol to do so consistently.
The update does not disable em dashes by default; rather, it refines the model’s responsiveness to user-defined writing rules. This change aligns with OpenAI’s broader goal of giving writers more stylistic control while maintaining natural, flexible output.
With the fix now in place, users can expect more predictable formatting, fewer unwanted punctuation quirks, and greater freedom to tailor ChatGPT’s tone and structure to match personal or professional needs.
