Privacy Policy

1. Who we are

The controller of your personal data is the business identified in the site footer (Verata, represented by Kim Se-jin). The privacy contact is the same as the contact shown at the end of this policy.

2. What we collect

Account: your name and email (via Google sign-in). Session settings: language pair, titles, and any glossary you add. Usage: minutes of translation, the providers used, cost, and timestamps. Transcripts: when a session opts in, and — during the current quality phase — for every session (see section 3). Technical logs (such as access IP and timestamps) needed to run and secure the service.

3. Audio and transcripts

Verata does not record or store conversation audio. Speech is streamed to our speech-recognition provider only to produce the live translation, then discarded. On-screen subtitles are ephemeral — held briefly in memory during the session and gone when it ends.

A transcript appears in your account when a session explicitly turns on transcript capture. Transcript lines are stored on the platform and automatically deleted after 30 days.

Quality phase (temporary): while Verata is being tuned, the text transcript of every session is stored so that translation errors can be found and corrected against real speech. Transcripts captured this way are not shown in your account, are used only to correct the service, follow the same 30-day deletion, and the whole set is deleted when the phase ends. Audio is still never recorded.

4. Why we process it

To provide the translation service, to bill usage, to provide support, and to keep the service secure.

5. Processing partners (subprocessors)

To run the service we entrust processing as follows. Speech recognition: Soniox (US). Translation: Anthropic Claude (US), Microsoft Azure (Japan), OpenAI (US). Voice synthesis: Microsoft Azure (Japan). Sign-in: Google (US). Email: Resend (US).

Each processor handles the stream in real time only to return a result and then discards it; under contract (and each provider's API terms) they do not use the entrusted data to train their models.

6. International transfers

Because some processors operate outside Korea, personal data is transferred abroad to the extent needed to deliver the service.

Recipients: Soniox, Anthropic, OpenAI, Google, Resend (US); Microsoft Azure (Japan). Data transferred: the live audio stream and text to be translated; account email (for sign-in and email delivery). Purpose: real-time recognition, translation and voice synthesis, sign-in, and notification email. When and how: transmitted over the network in real time at the moment of use. Retention: discarded immediately after processing (audio and translation streams are not stored).

You may refuse the international transfer; real-time translation and related features may then be unavailable.

7. Retention

Account data is kept while your account is active. As required by the Act on Consumer Protection in Electronic Commerce, records of contracts and withdrawals are kept for 5 years, records of payment and the supply of goods for 5 years, and records of consumer complaints or disputes for 3 years. Access logs are kept for 3 months under the Protection of Communications Secrets Act. Transcripts are automatically deleted after 30 days, whether opted in or captured during the quality phase.

8. Your rights

Under PIPA you may request access to, correction of, deletion of, or suspension of processing of your personal data, and you may withdraw consent. Contact us below to exercise these rights and we will act without undue delay.

9. Destruction

Personal data is destroyed without delay once its retention period has passed or its purpose has been fulfilled. Electronic files are deleted by a method that prevents recovery.