Following countless user complaints, OpenAI announced on Friday that it had solved one of the most talked-about problems in its artificial intelligence system: the frequent use of the em dash. Now, those who don't want to see the symbol in ChatGPT-generated texts can disable it in their personal settings.
For a considerable amount of time, ChatGPT users worldwide complained that, even when they explicitly asked the system to avoid using the em dash, the long horizontal line that appears between clauses, the system continued to insert it in almost every paragraph.
The symbol had become an unofficial hallmark of AI-generated writing.
In posts published on social media, OpenAI announced that the symbol can now be controlled through the user's personal settings.
"If you ask ChatGPT not to use em dashes, it finally does what it's supposed to," OpenAI CEO Sam Altman wrote, defining it as a "small but happy victory."
The em dash has caused an uproar among writers, editors, and educators over the past year, who claim it has become a tool for identifying AI-generated texts.
It appeared in academic papers, emails, online comments, customer service chats, advertisements, and LinkedIn posts. Many even claimed that its mere use suggested laziness on the part of the writer who relied on the AI software.
However, some defended the dash. Some argued that it is a "legitimate" and "elegant" punctuation mark that was used long before the age of artificial intelligence. On the other hand, the fact that large language models tended to use it frequently made it a suspicious sign, to the point that it was nicknamed "the ChatGPT dash."
For months, OpenAI attempted to address the issue, but users reported that the system continued to add the symbol even after explicit instructions to remove it. Now, with the new update, ChatGPT will follow users' personal instructions, though it will still use the em dash by default.
The meaning: every user can choose whether to keep the old writing style or abandon it completely.
ChatGPT custom instructions fix the em dash issue
The update was received enthusiastically online, and Altman's post received tens of thousands of likes. Many users thanked the company for listening to feedback and giving them back control over their writing style. Others, however, expressed nostalgia for the em dash, claiming it gave ChatGPT's texts a special literary flavor.
OpenAI noted that the new setting is now available to all users in both the ChatGPT Plus and the free versions, under the 'Custom Instructions' option in the personalization area. The company also published a post in which ChatGPT 'apologized' for 'ruining the em dash' and promised to operate under new rules going forward.