Introducing group chats in ChatGPT

OpenAI rolled out group chats inside ChatGPT, and the pitch is simple: instead of everyone copy-pasting the same prompt into their own separate window, you can now just invite people into one conversation and let the AI respond to the whole group at once. Think less "my chat with a bot" and more "a Slack channel where the bot is also a participant."

The feature is aimed squarely at the messy reality of how people actually use ChatGPT at work and in friend groups already — screenshotting answers, forwarding them, arguing about what the AI meant. OpenAI's framing leans on collaboration: teams planning a trip, roommates splitting a budget, coworkers drafting something together, all inside a shared thread where ChatGPT sees the full context of what everyone said, not just one person's slice of it.

That shared-context piece is the actual product shift here. Up to now, ChatGPT's memory and understanding lived inside a single user's conversation history. Group chats mean the model has to track multiple voices, figure out who's asking what, and respond usefully without getting confused about who said which thing three messages ago. That's a genuinely harder problem than single-user chat, and it puts ChatGPT closer to how messaging apps and collaborative docs already work, just with an AI sitting in the room.

It also nudges ChatGPT further from "personal assistant" toward "shared infrastructure," which is where OpenAI clearly wants it to sit alongside things like Slack, Notion, and Google Docs. Whether people actually want their AI assistant listening in on group conversations, versus firing off private one-on-one queries, is the open question. But the move fits a broader pattern of chatbots trying to become the default surface for getting things done with other people, not just alone at a desk.