TL;DR:
I’d like to suggest an optional feature that would allow two users’ personal ChatGPT assistants to communicate or collaborate with each other, but only with explicit permission from both users and with strict privacy controls.
The idea is not to share users’ private conversations, memories, or personal data with another person or their ChatGPT.
Instead, each user could explicitly authorize their own ChatGPT to exchange specific, limited information with another user’s ChatGPT for a particular purpose.
Example use case
Imagine two people who both use ChatGPT regularly and have personalized assistants that understand their individual preferences, communication styles, plans, or working methods.
One user could say:
“Ask Alex’s ChatGPT whether Saturday afternoon would work for our plan, and find an option that suits both of us.”
With permission from both sides, the two assistants could exchange only the information necessary to complete that task and return the result to their respective users.
This could also be useful for:
- coordinating plans between partners, friends or families;
- collaborative projects;
- organizing trips or events;
- resolving scheduling conflicts;
- comparing preferences when making a joint decision;
- playful interactions between users and their personalized assistants.
Privacy and consent would be essential
I imagine this working with very strict boundaries:
Both users must explicitly opt in.
Either user can revoke access at any time.
The assistants should not gain access to each other’s full memory or conversation history.
Each interaction could have a clearly defined permission scope, such as:
- one request only;
- a specific topic;
- a specific project;
- a limited period of time.
Before sensitive or personal information is shared, ChatGPT could ask the user for additional confirmation.
Users should also be able to see what information was exchanged between the assistants.
Why this would be different from shared chats or Projects
Shared spaces are useful when people deliberately work inside the same context.
This idea is different.
Each person would keep their own independent ChatGPT, private context, memory and relationship with their assistant. The assistants would simply have a controlled communication layer between them when their users authorize it.
In other words, it would be closer to:
“My ChatGPT can talk to your ChatGPT about this specific thing”
rather than:
“Let’s put both of us into the same shared AI workspace.”
I think this could create an entirely new kind of collaborative experience while still keeping individual accounts and personal context separate.
It could be useful for serious coordination, but also genuinely fun. Personalized assistants interacting on behalf of their users could produce some very interesting social and collaborative experiences.
I’d be curious whether something like this has been considered internally, and whether other users would find it useful.