Proposed feature: A temporary, secure connection between two ChatGPT users that enables natural, real-time multilingual conversation.
Imagine two people meeting in person who do not speak the same language—for example, a French-speaking Canadian and a Japanese-speaking person. Both already have ChatGPT on their phones.
Instead of using one phone as a conventional translation device and taking turns speaking into it, the two users could temporarily pair their ChatGPT sessions by scanning a QR code, using NFC proximity, or another secure pairing mechanism.
Once paired, each person would continue using their own phone, their own ChatGPT account, their preferred voice, and their preferred language.
For example:
- User A speaks French → User B hears Japanese.
- User B speaks Japanese → User A hears French.
- ChatGPT handles the interpretation continuously and automatically, without requiring either person to press a translation button after every sentence.
The objective would be to make the technology progressively disappear from the interaction. Rather than feeling like two people operating a translation application, it should begin to feel like two people simply having a conversation.
Account-to-account communication
The important distinction from existing live translation systems is that this would not simply be one device listening to two languages.
It would establish a temporary communication channel between two independent ChatGPT sessions. Each instance could optimise the experience for its own user while exchanging only the information necessary to interpret the conversation.
This architecture could eventually allow each user’s ChatGPT to preserve individual preferences such as language, voice, speaking speed, accessibility requirements, and translation style without exposing private account information to the other participant.
Privacy by design
Pairing should be temporary and explicitly authorised by both users.
Neither participant should gain access to the other person’s ChatGPT history, memories, account information, contacts, or personal data. The connection should share only what is required for the live conversation.
When either participant selects End Conversation, the temporary connection would be terminated.
An optional mode could also allow users to choose whether a transcript is retained or whether the entire interpreted conversation remains ephemeral.
A possible future extension
This becomes particularly interesting when combined with earbuds or AI glasses.
Imagine arriving in Japan, meeting someone, briefly pairing your ChatGPT sessions, and then putting your phones back into your pockets. You speak French naturally and hear the other person in French through your earbuds; the Japanese participant speaks Japanese and hears you in Japanese through theirs.
With AI glasses, translated speech could additionally appear as discreet subtitles.
At that point, ChatGPT would no longer function primarily as a translation application. It would become a real-time linguistic bridge between two people.
Why this could matter
The same protocol could potentially work anywhere two people need to communicate across a language barrier: travel, international conferences, education, healthcare, customer service, immigration services, business meetings, or simply two people meeting socially.
The fundamental feature request is therefore:
Allow two ChatGPT users to temporarily pair their sessions and create a private, real-time, bidirectional interpretation channel between their respective languages—while each person retains their own device, account, voice interface, preferences, and privacy.
This could turn live translation from a device-centred interaction into a genuinely person-to-person conversation mediated by AI.