OpenAI and X are making big moves to reshape how we chat online, each rolling out updates that aim to improve how users communicate and collaborate with AI and each other.
TLDR:
- OpenAI has finally fixed ChatGPT’s overuse of em dashes, a major annoyance for writers and editors.
- ChatGPT now respects custom style instructions more reliably across conversations.
- X, formerly Twitter, unveiled “Chat,” its new messaging system with encrypted group chats, voice, and video calls.
- Both updates point to a broader trend of platforms enhancing user control and privacy in digital conversations.
What Happened?
OpenAI has quietly resolved one of ChatGPT’s quirks that drove editors mad: its obsession with em dashes. CEO Sam Altman confirmed the fix on X, calling it a “small-but-happy win” for users. Meanwhile, X introduced a complete overhaul of its direct messaging system with a new platform called Chat. It includes end-to-end encryption, group support, media sharing, and video calls, bringing X in line with other major messaging apps.
OpenAI Tackles the Em Dash Problem
For nearly two years, ChatGPT users struggled with the AI’s inability to avoid using em dashes, even when explicitly instructed not to. Writers and editors frequently voiced their frustration, turning the issue into a long-running internet meme.
OpenAI CEO Sam Altman announced the resolution via a post on X, saying, “Small-but-happy win: If you tell ChatGPT not to use em dashes in your custom instructions, it finally does what it’s supposed to do!”
Small-but-happy win:
— Sam Altman (@sama) November 14, 2025
If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do!
Testing shows that the model now responds properly when asked to avoid the punctuation and remembers the preference through multiple conversations, thanks to improved instruction memory. Content creators have welcomed the change, noting that it reflects a broader effort by OpenAI to respect detailed style requests.
Still, the rollout appears to be gradual. Some users report the occasional dash still slipping through, and a few pointed out that more pressing issues remain, like factual accuracy in AI responses.
OpenAI Adds Group Chat Functionality
In a more significant move, OpenAI also introduced Group Chat functionality for ChatGPT. This feature allows up to twenty users to collaborate in a shared chat space, where ChatGPT can assist everyone by brainstorming ideas, answering questions, or helping with group tasks.
This new feature is currently available in select countries, including Japan, New Zealand, South Korea, and Taiwan. By starting with a limited rollout, OpenAI hopes to gather user feedback before expanding the tool globally.
Group Chat is part of a larger trend by OpenAI to shift ChatGPT from a solo experience into a more social and collaborative platform. Recent launches like Sora, a short-video creation tool, and Atlas, a browser for Mac users, also support this vision.
X Launches “Chat” with Encryption and Video Calls
Not to be outdone, X has introduced a comprehensive update to its messaging system, replacing traditional DMs with a more feature-rich platform named Chat.
The update brings:
- Voice and video calling
- File sharing
- Message editing and deletion
- End-to-end encryption for messages and media
- Screenshot notifications
These new features are rolling out on iOS and the web, with Android support coming soon. A voice memo tool is also in development.
Say hello to Chat – all-new secure messaging on X.
— Chat (@chat) November 14, 2025
• end-to-end encrypted chats and file sharing
• edit, delete, or make messages disappear
• block screenshots and get notified of attempts
• no ads. no tracking. total privacy. pic.twitter.com/7dmDEDkYvO
Voice memos are returning soon – what else would you love to see in Chat?
— Chat (@chat) November 14, 2025
X had previously launched encrypted DMs earlier this year but paused them to fix some limitations. According to the company’s updated help documents, group messages and shared media can now be encrypted. However, metadata such as recipient information is not protected, and X acknowledges that its current system does not guard against man-in-the-middle attacks.
The platform notes that if conversations are compromised by a malicious insider or a legal request, users would not be notified, highlighting the limitations of the encryption. X says it is developing ways for users to verify encrypted chat authenticity in the future.
Daily Research News Takeaway
I really like what I’m seeing here. These updates show that OpenAI and X are finally listening to user feedback and fixing real problems. Whether it’s the overuse of a pesky punctuation mark or the lack of secure messaging, these changes are about improving everyday communication. Group chats in ChatGPT make it more useful in team settings, and X’s move toward privacy-first messaging is a much-needed step in the social media space. While both platforms still have work to do, this feels like real progress.
