OpenAI Fixes ChatGPT's Em-Dash Obsession with Customization

As of 2025-11-15T18:53:06+00:00, OpenAI has stepped in to address the ongoing issue of ChatGPT's overuse of em-dashes in its generated writing. According to CEO Sam Altman, users can now customize their settings to prevent the chatbot from using em-dashes going forward.

The em-dash fixation has proven controversial since it first started appearing in ChatGPT's responses. Research from The Washington Post showed that at one point this year, almost half of ChatGPT's output included the punctuation mark, a nearly fivefold increase compared to 2024. Many ChatGPT users had taken to social media to complain about the difficulty in getting the chatbot to stop using em-dashes, despite trying various prompts.

The widespread use of em-dashes in ChatGPT's writing has put some writers in a difficult position, as the punctuation mark has become associated with AI-generated content rather than human-written work. Some users have even resorted to avoiding em-dashes altogether to prevent unfair accusations.

While OpenAI has not explained the exact reason behind ChatGPT's fondness for em-dashes, researchers have proposed a few theories. One suggests that the em-dash was very common in 19th-century books, which are often used as training data for large language models like ChatGPT. Another theory points to the widespread use of em-dashes on Medium blogs, another common source of OpenAI's training data.

With the new customization options, ChatGPT users can now take control of their writing experience and ensure the chatbot no longer overuses the em-dash. However, the jury is still out on whether this fix will work for everyone, as Ars Technica reported a response from ChatGPT that included an em-dash despite being instructed never to use it.