OpenAI Pilots Group Chats in ChatGPT: A Step Toward Collaborative AI Interaction

 

OpenAI has quietly launched a limited pilot for group conversations within ChatGPT, marking a significant evolution in how users interact with AI in shared, multi-user environments. The feature enables real-time collaboration among friends, family, or colleagues, with ChatGPT functioning not as a separate tool, but as an integrated, context-aware participant in the discussion. The pilot is currently restricted to users in Japan, New Zealand, South Korea, and Taiwan, and is accessible to Free, Plus, and Team plan subscribers on both web and mobile platforms.

To initiate a group conversation, users tap the people icon in the top-right corner of a new or existing chat. Starting from an existing one-on-one conversation creates a new group without carrying over prior message history. Up to 20 participants can be added directly by username or via a shareable link. Anyone with the link can invite others, making group growth organic and flexible. Each group receives a short profile and appears in a clearly labeled section of the sidebar for easy navigation.

Once formed, the interface mirrors familiar messaging apps, but with AI deeply woven into the flow. Powered by GPT-5.1 Auto, ChatGPT is designed to mimic human conversational etiquette:

•  It observes the discussion and responds only when relevant.
•  Users can summon it explicitly by typing @ChatGPT, ensuring control over when AI input is desired.
•  It supports full multimodal capabilities, including real-time web search, image generation (including personalized visuals using participants’ profile photos, if provided), file uploads and analysis, voice dictation, and emoji reactions.

A key usability detail is that only ChatGPT’s responses count toward hourly usage limits—all human messages are exempt. This design encourages free-flowing discussion without fear of quota depletion.

OpenAI has implemented layered privacy controls:

•  One-on-one chats and personal memory remain completely isolated from group interactions.
•  If any participant is under 18, content filtering applies to the entire group, with enhanced parental oversight options.
•  Groups are invitation-only during the pilot, preventing public discovery.
•  Users can mute, remove others (except the creator), or leave at any time. The group creator cannot be removed but can exit voluntarily.

The feature lends itself to a wide range of collaborative scenarios:

• Travel planning: A family brainstorms a multi-city itinerary, with ChatGPT suggesting flights, stays, and hidden gems.
• Home projects: Friends design a kitchen remodel, with the AI generating layout options and material cost estimates.
• Team work: Colleagues outline a report, with ChatGPT structuring content, citing sources, and refining drafts.
• Social recommendations: A dinner group debates restaurants, with the AI filtering by cuisine, budget, and real-time availability.

The pilot operates under specific technical and regional constraints:

• Underlying model: GPT-5.1 Auto, optimized for low-latency, context-aware responses.
• Maximum participants: 20 per group.
• Supported platforms: Web, iOS, and Android.
• Eligible plans: Free, Plus, and Team.
• Pilot regions: Japan, New Zealand, South Korea, and Taiwan.
• Global rollout: No timeline announced; expansion depends on pilot feedback.

OpenAI positions this pilot as a small first step toward a more shared experience with AI. The company has been exploring social dynamics in AI interaction. This group chat feature suggests a broader vision: transforming ChatGPT from a personal productivity tool into a collaborative social intelligence layer. By learning from real group dynamics—turn-taking, relevance, humor, and restraint—OpenAI aims to train models that participate more naturally in human social contexts.

Pilot users are encouraged to submit feedback directly through the app. OpenAI has indicated that insights from this phase will shape:

• Expansion to additional countries.
• Potential integration with enterprise workflows.
• Enhanced moderation and customization tools.
• Possible public or workspace-based group discovery post-pilot.

In summary, OpenAI’s group chat pilot represents a deliberate, user-centered experiment in multi-person AI collaboration. By starting small, regionally focused, and feedback-driven, the company is gathering critical data on how AI can enhance—not interrupt—human group interaction. For now, users in the pilot regions can begin exploring this new paradigm of shared intelligence.