OpenAI has begun piloting group chats in ChatGPT across four Asia-Pacific countries—Japan, New Zealand, South Korea and Taiwan—turning its flagship AI assistant into a shared space where multiple people and an AI can collaborate inside the same conversation.
What OpenAI Is Rolling Out
OpenAI is introducing a new group chat feature inside ChatGPT that allows several human users to participate in a single thread with the AI, instead of the traditional one‑to‑one chat. The pilot is live on the web, iOS and Android apps for logged‑in users across all major plans—Free, Go, Plus and Pro—in the four selected countries.
Each group can include up to 20 participants alongside ChatGPT, making it suitable for anything from family planning and study groups to team brainstorming sessions. OpenAI describes it as an early step toward “shared experiences” in ChatGPT, moving the product beyond a solitary assistant into a collaborative communication and productivity tool.
The Four Pilot Countries
The group chat experiment is limited, for now, to Japan, New Zealand, South Korea and Taiwan. These markets give OpenAI a mix of highly connected, mobile‑first users and advanced regulatory environments, useful for testing new social features that mix messaging, AI assistance and safety controls.
The feature is available only inside OpenAI’s own ChatGPT apps and website in these regions, not yet integrated into external messengers or productivity suites. OpenAI says it plans to expand to more countries and potentially more platforms after gathering early usage data and feedback.
How ChatGPT Group Chats Work
Users can start a group chat by tapping the people icon in any new or existing conversation, after which an invite link can be shared to bring others into the room. When an existing one‑to‑one chat is turned into a group, ChatGPT creates a copy for the group, preserving the original private thread separately.
Within a group, participants set up lightweight profiles with a name, username and photo so everyone can see who is speaking. Group chats are organized in their own section in the ChatGPT sidebar, making them easier to locate and manage alongside individual conversations.
AI That Behaves Like a Participant
A key design choice is that ChatGPT behaves more like a human participant than a bot that responds to every single message. The system follows the flow of conversation and decides when to respond or stay quiet, based on context and the group’s activity.
Users can explicitly call on the AI when needed—such as by mentioning “ChatGPT” in the thread or using an @‑style mention—if they want a guaranteed response. The assistant can react with emojis to messages and, when prompted, use group members’ profile photos to generate fun, personalized images for everyone in the chat.
Powered by Newer Models and Tools
Behind the scenes, group chats run on OpenAI’s latest GPT‑5.1‑based “Auto” experience, which can decide when and how to respond within a conversation. Within these rooms, ChatGPT has access to search, file and image uploads, image generation, and dictation, allowing it to summarize documents, analyze shared files and generate content in real time for the entire group.
Importantly, only ChatGPT’s own messages count toward usage limits, not the messages exchanged between human participants. That structure is meant to encourage free discussion among members without worrying that every human message will exhaust plan quotas.
Collaboration and Use Cases
OpenAI pitches group chats as a tool for joint planning, decision‑making and creative collaboration, with the AI acting as an assistant, mediator or “third brain” that everyone can tap into. Official use‑case examples range from coordinating trips or dinner plans to running study groups, home‑design discussions and project rooms where teammates drop links, files and images for ChatGPT to summarize and build upon.
For workplace scenarios, up to 20 colleagues can gather in a single conversation, ask ChatGPT to generate drafts, compare ideas or synthesize the overall discussion into action points and timelines. For families and friends, the company highlights casual planning, games and creative prompts as ways to turn ChatGPT into a shared activity rather than an individual tool.
Privacy, Safety and Controls
OpenAI is emphasizing privacy and safety safeguards as it moves ChatGPT into a more social, multi‑user context. Group conversations are handled separately from one‑to‑one chats: when more than one human is present, ChatGPT’s personal memory feature is automatically turned off, and no new memories are created from those group discussions.
Account‑level custom instructions and private chat histories are not exposed inside group rooms, reducing the chance that personal context spills into shared threads unexpectedly. Participants can see who is in the group, leave at any time, remove others (depending on permissions), reset or delete invite links, and delete their own messages for everyone in the chat.
Protections for Younger Users
Because group chats can involve friends, classmates and families, the feature includes specific protections for teenagers. If an under‑18 user is present in a group, ChatGPT applies stricter filters that reduce sensitive or adult content for the entire conversation.
Parents and guardians also have the option to disable group chat entirely for teen accounts, as part of OpenAI’s broader parental control framework. These settings aim to address concerns that multi‑user AI spaces could expose minors to inappropriate material or interactions.
Strategic Stakes and Competitive Context
The move into group chats comes as generative AI products race to become embedded in everyday communication tools, not just standalone bots. Microsoft has already experimented with multi‑person collaboration in its Copilot ecosystem, and OpenAI’s pilot positions ChatGPT more directly as a competitor to traditional messengers and collaboration platforms that are layering AI on top.
By starting in four countries and limiting access to its own apps, OpenAI is effectively running a controlled test of how people behave when AI sits in the middle of group dynamics—planning, debating and sometimes arbitrating disagreements. The company signals that this is a “small first step” toward a broader vision in which ChatGPT is not just a private assistant, but a shared environment for teams, families and communities to think together with AI.
What Comes Next
For now, users outside Japan, New Zealand, South Korea and Taiwan will have to wait to try group chats. OpenAI plans to adjust features such as when ChatGPT speaks up, how notifications work and what moderation tools are needed, based on early feedback from the pilot markets.
If uptake is strong and safety systems hold, the company is expected to roll the feature out to more regions and potentially deepen integrations with productivity suites and external platforms. That expansion would mark another step in turning ChatGPT from a single‑user Q&A engine into a central hub for collaborative work and conversation in the AI era.
