OpenAI brings group chats to ChatGPT

Planning a weekend trip with friends usually involves one person aggressively copying itineraries between WhatsApp and notes apps, trying to get everyone to agree.

OpenAI wants to cut out that middleman. The company just launched group chats inside ChatGPT. After kicking off a pilot in Japan, New Zealand, South Korea, and Taiwan on November 13, OpenAI announced it is expanding the feature globally to all logged-in Free, Go, Plus, and Pro users.

An AI that knows when to shut up

Starting a group chat is straightforward. You tap the people icon in the top right corner of any conversation to create an invite link for up to 20 people. If you invite someone into an ongoing chat, ChatGPT creates a duplicate copy as a new shared space, keeping your original chat completely separate.

Each member can set a short profile with a display name and photo. The chats live in a dedicated, clearly labeled section in the sidebar.

Under the hood, responses are handled by GPT-5.1 Auto, which selects the model based on the prompt and the specific user's tier. Features like web search, file uploads, image generation, and dictation are all supported.

What is particularly neat are ChatGPT's new social behaviors. Instead of chiming in on every single text, it reads the room. The AI tracks the conversation flow and decides when to answer and when to stay quiet, though you can always explicitly mention ChatGPT to force a reply. It can also react with emojis and reference profile photos to generate inside-joke images.

Crucially, chatting between humans does not burn your rate limits. The usage limits only kick in when ChatGPT actually posts, and it only counts against the quota of the specific person the AI is responding to.

Privacy walls and clean modes

OpenAI also tackled privacy head-on. Your personal ChatGPT memory is not accessible in group chats, and the model will not store new personal memories from group discussions.

Joining requires accepting an invite, and anyone can leave whenever they want. Group members can remove others, but only the group creator can remove themselves by leaving. If an under-18 user joins, ChatGPT automatically tones down sensitive content for everyone in that room, and parents can disable group chats altogether via parental controls.

Why it matters

We think this is a clear signal that OpenAI wants ChatGPT to be more than a solo utility tool. It is pushing hard into collaborative workflows.

For everyday users, the immediate pitch makes a lot of sense. Planning a vacation itinerary, settling group debates, or brainstorming school projects in real time with an impartial assistant saves a massive amount of cross-app screenshot pasting. It is definitely worth trying for small task-driven crews, though nobody is going to ditch iMessage or WhatsApp for casual chatter anytime soon.

For the industry, OpenAI is cutting into territory traditionally held by Slack and Discord bots. Instead of users hooking up custom API integrations into workplace messengers, OpenAI is pulling the multi-user canvas directly into its own product, keeping eye-balls and engagement inside its walled garden.

OpenAI handled the guardrails smartly. Making human messages free from rate limits keeps the room relaxed, and sandboxing personal memories prevents embarrassing privacy leaks. But the tricky part will be conversation dynamics: balance is everything. If the AI talks too little, people will forget it is there; if it talks too much, it becomes that annoying acquaintance ruining the group vibe.