New Delhi, Nov. 14 -- When ChatGPT first launched in late 2022, the big debate was how one would differentiate between AI-written and human-written content. Since then, users have found that AI tools leave a lot of tells that clearly differentiate them from human content, and perhaps the most prominent among them is the excessive use of em dashes.

The problem that occurred with ChatGPT and other AI tools continued to persist even when users told the AI specifically not to use em dashes. However, OpenAI CEO Sam Altman now says that the company has fixed the problem and that if users tell ChatGPT not to use dashes, it will follow those instructions.

"Small-but-happy win: If you tell ChatGPT not to use em dashes in your custom instructions...