Archive — US — techcrunch
OpenAI says it’s fixed ChatGPT’s em dash problem
OpenAI has addressed the issue of ChatGPT's overuse of em dashes. Users can now instruct the AI to avoid using this punctuation in its outputs, giving them greater control. The em dash, criticized for being a hallmark of AI-generated text, can lead to negative perceptions of writing quality. CEO Sam Altman announced this fix on social media, marking it as a positive enhancement for user customization.
Key points
- OpenAI has fixed the ChatGPT issue with em dashes, allowing users to customize output.
- The em dash has been criticized as a telltale sign of AI-generated text.
- Some writers defend the em dash as a legitimate punctuation mark used long before AI.
- Users previously struggled to prevent ChatGPT from using em dashes despite requests.
- Sam Altman confirmed the update via a social media post, celebrating user influence on AI outputs.
Why it matters
- The trend of AI-generated text raising concerns about writing quality reflects broader discussions in 2025 about AI ethics and authenticity.
- Critiques of AI outputs are part of a larger narrative on how technology shapes communication styles.
- The fix aligns with the rise of personalization in AI tools, enabling users to tailor experiences based on preferences.
In summary, OpenAI's adjustment to ChatGPT's handling of em dashes enhances user control over AI outputs, demonstrating the importance of user feedback in improving AI interactions as technology continues to evolve.