← All articles

Privacy

Is it safe to use a meeting translator extension?

August 1, 2026

A translator listens to your meeting, so it is fair to ask where the audio goes. Here is how these tools work and what to check before you trust one.

Two different questions

Separate two things. First: does a bot join the call? A bot that joins is visible to everyone and often shows up in the participant list and recordings. An overlay tool, by contrast, runs only on your side — nothing is added to the meeting.

Second: where does the audio go to become text? That is the part worth checking, because most live translators use a cloud service to turn speech into text.

How speech becomes translated text

To caption a call, a tool captures the audio, sends it to a speech-to-text service to get words, then to a translation service. This is normal and how near-instant captions are possible. The questions that matter are whether that audio is stored, whether your identity is attached to it, and whether the connection is encrypted.

What to check before you trust one

• No bot joins the call — captions appear only on your screen.
• Audio is processed in real time and not stored.
• A clear privacy policy that names what is sent and to whom.
• Minimal permissions — it should not need your browsing history or unrelated sites.
• You control when it runs — nothing is captured until you press Start.

How Trippi handles it

Trippi adds no bot: the caption overlay is drawn only on your screen. It captures the meeting tab’s audio while a session is active, sends it to speech and translation providers only to return captions, and does not store it. Provider keys stay on Trippi’s server, not in the page, and traffic is over HTTPS. Nothing is captured until you press Start, and it stops the moment you press Stop or close the tab. See the privacy policy.

One practical note

Captioning a call is about understanding it, not recording it — but rules on consent and recording vary by place and organization. Follow the ones that apply to you, and remember you decide when the tool is on.

What the permissions actually mean

When you install a captioning extension, Chrome shows a permission prompt that sounds alarming and is rarely explained. “Read and change all your data on meet.google.com” is what the browser says whenever an extension can run code on a page — for a captioning tool, that is what draws the overlay and reads the audio element. It is a broad phrase for a narrow job, but the breadth is real, which is why it matters who wrote the extension.

The permission worth examining more closely is tab capture: the ability to take the audio of a tab. That is the core capability of any tool in this category, and a tool that did not need it could not caption anything. What differs between products is not whether they capture the audio but what happens to it next.

Questions worth asking any vendor

• Is audio stored after the captions are produced? If yes, where and for how long?
• Which third parties process it — the speech recognition and translation providers — and are they named anywhere?
• Are captions or transcripts kept on a server, or do they exist only on your screen?
• Is your audio used to train models?
• Is there usage telemetry, what does it contain, and can it be switched off?
• Is there a privacy policy that answers all of the above in specifics rather than in adjectives?

Consent, policy and the part that is not technical

There is a difference between captioning a call for yourself and recording it, and it is worth keeping the two apart. Captions that appear on your screen and are not stored are not a recording. That said, in several jurisdictions the rules around capturing what is said on a call are strict, and many employers have their own policy regardless of what the law requires.

The practical advice is boring and correct: if it is a work call, check your company policy before adding any tool to it, and if you are unsure whether captioning counts, ask. This is not legal advice — it is the same caution you would apply to any tool that touches a conversation.

How Trippi answers those questions

Audio is captured from the meeting tab, sent to speech recognition and translation providers to produce the caption, and not stored. Nothing joins the meeting, so no other participant is affected and nobody is recorded.

The text is a separate question, and worth being precise about. If you leave the transcript feature on, what was said is kept in your browser’s local storage on your own machine — not on a server. You can delete a meeting, delete all of them, or switch the feature off in the settings. Two things send a copy to ours, and both are named in the settings and switchable there: the email with the transcript, which goes to your own account address after a call, and a summary, which is built only when you press the button for it. Neither is stored on our side, and neither can be pointed at anybody else’s address.

The extension reports anonymous usage events — whether captions started, on which platform, why they failed, how long a session ran — so that breakage can be found and fixed. It never reports captions, meeting names or the pages you visit, and the reporting can be turned off in the extension’s settings. The full detail is in the privacy policy, which is worth reading rather than taking on trust — including for tools other than this one.