OpenAI lets users control em dash usage in ChatGPT

OpenAI CEO Sam Altman revealed on X on Friday that ChatGPT will now follow users’ custom instructions regarding em dashes. Previously, even if users asked the AI not to use em dashes, the chatbot often ignored the request.

It gives users more control over ChatGPT's writing style, making human or AI writing even harder to distinguish.

That's a significant update because excessive em dash usage has started being treated by many internet users as a clue that a piece of text might be AI-generated.

While the presence of em dashes does not necessarily indicate AI authorship, people have grown suspicious of texts with heavy use of the punctuation.

It isn't entirely clear why generative AI models use so many em dashes. Large language models (LLMs), such as ChatGPT, are trained on enormous volumes of content, from books and academic papers to forum posts and online articles.

How to stop ChatGPT from using Em Dashes

1. Open ChatGPT on your browser, mobile app, or desktop.
2. On desktop: Click your profile icon in the bottom left corner, then select Settings > Custom Instructions.
3. On mobile: Tap the three dots or menu icon, then go to Settings > Custom Instructions.
4. Under the box of custom instructions, write:
5. “Do not use em dashes in your responses. Use other punctuation like commas, colons, or parentheses instead.”
6. Click Save or Done to apply your instructions.

The prevalence of em dashes in training material might explain why LLMs naturally insert them, and it would appear that AI trainers haven't flagged this as an issue to avoid.