Privacy Policy

Template — have a lawyer review before you operate this publicly.

What this deployment processes

• Connection data — network address, approximate country, browser and device type, and a random client identifier generated in your browser.
• Session metadata — when a conversation started and ended, and which anonymous identifiers were involved.
• Moderation signals — text and images are screened in transit; flagged items are recorded with a reason code.
• Local preferences — filters, interests, recent calls, blocked identifiers, and your age confirmation are stored in your own browser, not on the server.

What is never collected

No video. No real name, phone number, or date of birth. No precise location. No recording of a call is ever made or stored. With live captions switched off, your voice travels directly between the two participants' browsers and never passes through the server at all.

Voice pitch analysis

When the optional gender preference filter is enabled, short segments of your own microphone input are analysed inside your browser to estimate voice pitch. Only the resulting one-word estimate is sent with a match request. That audio never leaves your device.

Live captions and translation

This one is an exception to everything above, which is why it is off by default and has its own switch.

When you turn captions on, short clips of your microphone — a few seconds each — are sent to the server and passed to a speech-to-text provider so they can be turned into words. If your conversation partner has asked for translation, that text is then sent to a translation provider. The audio clips are held in memory only for as long as the request takes, are never written to disk, and are discarded immediately afterwards. Neither the audio nor the resulting text is stored.

Captions apply only to your own microphone, and only while you have the setting enabled. Turning it off stops the clips immediately. If you would rather no audio ever reach the server, leave captions off — the rest of the service works exactly as before.

Spoken-word moderation

While captions are enabled, the transcribed text is also checked against the same content rules applied to typed messages, and a line that breaks them is not passed on to the other person. Repeated spoken violations lead to a temporary suspension. Only the reason code and a timestamp are recorded — never the words themselves, and never the audio.

Retention

• Session metadata — kept only while the service is running; nothing is written to disk in this build.
• Moderation flags — retained in memory, capped to the most recent entries.
• Random text messages — relayed in memory, never stored.
• Anything in your browser's local storage — until you clear it.

If you connect a database to this deployment, define concrete retention periods here and enforce them in code.

Third parties

This build uses public STUN servers to help two browsers find each other, and optionally an image moderation provider if an API key is configured. If you add analytics, advertising, hosting, email, or payment providers, list every one of them here along with the country they process data in.

Your rights

Depending on where you live you may have the right to access, correct, delete, restrict, or port your personal data, to object to processing, and to withdraw consent. Because the service requires no account, the practical way to erase everything associated with you is to clear this site's data in your browser. For anything else, contact the operator.

Children

The service is not directed at and may not be used by anyone under 18. Data suggesting an underage user is used solely to remove that user from the platform.

Contact

Insert the data controller's legal name, address, and contact address for this deployment.