OpenAI has started pilot testing a new group chat feature for ChatGPT in Japan, New Zealand, South Korea and Taiwan. The option lets users create shared conversations where ChatGPT participates alongside multiple people.
Within these chats, the assistant can help plan trips, suggest renovation ideas or find restaurants that match everyone’s preferences. It can also support school or work projects by structuring reports from the notes and files shared in the group.
Users can begin a group chat by tapping the people icon in the upper-right corner of the screen while inside a new or existing conversation. If the feature is started from an existing chat, ChatGPT creates a fresh thread, so previous history remains private. Participants can be added manually or through a link, and up to 20 people can join. Each person must create a profile that includes a name, username and photo.
ChatGPT group chats support the same tools found in one-on-one conversations, such as file uploads, image generation and search. Anyone with the invite link can bring others into the chat unless the creator restricts this. Members can mute or remove others at any time, though the creator cannot be removed. When someone under 18 is part of the group, ChatGPT automatically limits sensitive content for all participants.
The conversations run on GPT-5.1 Auto, which selects the most suitable model depending on the prompt and the user’s plan. According to OpenAI, the assistant is trained to follow natural group-conversation patterns, choosing when to stay quiet and when to respond. Users can also call the assistant directly by mentioning “ChatGPT”.
OpenAI says group chats are separate from private chats and do not share personal ChatGPT memory across participants. The company is exploring additional user controls as the feature evolves. Parental control settings allow guardians to disable group chat access entirely for younger users.
For now, the pilot is limited to four regions, and it is available on both free and paid ChatGPT plans. OpenAI plans to refine the feature based on feedback from early testers before expanding availability to more countries.