OpenAI CEO Sam Altman made a small but interesting announcement today: ChatGPT should now follow user instructions to avoid using em-dashes. He called it a “small- but-happy win.” But what seemed like a simple update quickly turned into a lively online conversation one filled with jokes, confusion, and plenty of screenshots suggesting the fix might not be fully working yet. The post didn’t go unnoticed. It quickly crossed 274,000 views and gathered hundreds of comments and reactions across X. A tiny punctuation update, a big online debate Altman’s message was light-hearted, but it opened the door to a surprisingly passionate discussion about punctuation. Many users have long complained about ChatGPT’s love for em dashes, saying that the bot uses them even when asked not to. So, when Altman claimed the issue was fixed, people rushed to test it out for themselves. Soon, the comments were filled with jokes, memes, and personal experiences. Some users celebrated the update. Others were skeptical. A few even poked fun at themselves for caring so much about punctuation. Screenshots suggest it’s still not perfect Even though Altman said the problem had been resolved, not everyone agreed. Several users posted screenshots showing ChatGPT still slipping an em-dash into replies even after clear instructions not to. These posts quickly spread as more people ran their own mini-experiments, checking whether the chatbot actually followed the new rule. My test: Did ChatGPT really stop using Em-dashes? To see what was really going on, I tried it myself. I told ChatGPT: “Please do not use em dashes.” The reply I got is,“ Got it — I’ll avoid using em dashes from now on.” Yes, you saw that right ChatGPT used the very punctuation mark it promised to avoid. This echoed the same issue many users were reporting: the update may not be fully consistent yet. Users confused about how to enable the setting Another part of the debate came from people who didn’t even know how to activate the no ’em-dash’ instruction. A user with the handle @sandy_carter asked how to turn it on because they had been trying repeatedly but couldn’t get it to work. This question gained attention, with others stepping in to discuss whether the feature was automatic or needed a specific setting. But the confusion grew when even those who did instruct the model clearly still saw occasional em-dashes slipping in sometimes even in responses where ChatGPT was explaining that it would avoid them. The mixed reactions showed that even tiny changes in an AI system can spark big discussions especially when millions of people rely on it every day.
The post appeared first on .
