Security and data handling
Last updated November 13, 2026
This page answers, in plain facts, what happens to a recording you give JustTranscribe: where it is stored, for how long, which other companies process it, who can see it and how you remove it. It also says what we do not have, so nobody has to guess.
It is a summary of how the service works today. The privacy policy is the legal document; where the two differ, the policy governs.
At a glance
- Where data is stored
- Amazon Web Services, Tokyo region (ap-northeast-1): the application, the database and your stored media.
- In transit
- HTTPS (TLS) on every page and API call.
- Who can see a transcript
- Only the account that made it — unless you create a public share link, which you can revoke.
- Video files
- Removed 3 days after processing.
- Audio files
- Kept while the account is in use; removed after 7 days without a visit. With Pro, media is kept while the subscription is active.
- Transcripts, summaries, translations
- Kept until you delete them.
- Deletion
- Delete any transcript, or the whole account, yourself at any time; the stored media goes with it.
- Model training
- We do not train models on your recordings or transcripts.
- Payments
- Handled by Stripe. Card numbers never reach our servers.
- Certifications
- None. No SOC 2 or ISO 27001 report, no HIPAA agreement.
What happens to a recording
A file you upload goes over HTTPS to our server on AWS. The audio is sent to a speech-recognition provider (xAI or OpenAI, chosen by the language) and comes back as text; that text is sent to a language-model provider (DeepSeek, with Anthropic as the backup) to write the summary, and again if you ask for a translation or a cleanup. The transcript and those results are saved to your account.
When you paste a public link instead, our server downloads the audio itself. If the direct download fails, a video-fetch provider (Apify, LamaTok or HikerAPI) receives the public link — never a file you uploaded.
Who processes your data
These companies receive data to run the service, each only what its job needs:
- xAI and OpenAI — the audio, to transcribe it.
- DeepSeek and Anthropic — the transcript text, for the summary, translation and cleanup.
- Deepgram — the audio of one recording, only when you turn on Speakers for a language where labels are not automatic.
- Apify, LamaTok, HikerAPI — a public video link, when our own download fails.
- Amazon Web Services — hosting and storage.
- Stripe — your payment details (directly from you), plus your email and country, if you subscribe to Pro.
- Google — sign-in, when you choose "Sign in with Google".
- Cloudflare — bot protection on sign-up.
- Resend — your email address, to deliver the emails the service sends you.
- PostHog — cookieless product analytics: which screens and buttons are used, never the content of a transcript.
These providers handle the data under their own API terms. We have not negotiated custom zero-retention or data-residency agreements with them, so we do not promise more than their standard terms do.
How long things are kept
- Uploaded or fetched video: removed 3 days after processing.
- Audio: kept while you use the account, so you can replay it next to the text; removed once the account has gone 7 days without a visit. The transcript stays.
- With an active Pro subscription, the audio and video of your finished transcripts are not removed by those two rules.
- Transcripts, summaries, translations and exports: kept until you delete them or the account.
- Deleting a transcript deletes its media. Deleting the account (Profile → Delete account) deletes everything in it.
Access and accounts
- Transcripts are private to the account by default. Every request is checked against the owner; another account gets "not found".
- A share link is opt-in, uses a long random token that cannot be guessed, shows a read-only page, and stops working the moment you revoke it.
- Sign in with Google, or with email and a password stored only as a salted hash.
- API keys are shown once and stored only as a hash; you can revoke a key at any time. Connections from Claude, ChatGPT and other apps use OAuth and appear in the same list.
- Access to the production server and database is limited to the people who operate the service.
What we do not have
Stated plainly, because these are the questions a business should ask:
- No SOC 2 or ISO 27001 audit.
- No HIPAA business associate agreement — do not upload protected health information.
- No single sign-on (SAML), no team workspaces with roles, no audit log for customers.
- No choice of storage region: data is stored in Japan and processed by providers in the United States and elsewhere.
- No human review of transcripts: the output is automatic.
If your recordings fall under rules that need any of the above, this is not the right tool for them yet. For everyday work — interviews, classes, meetings, public videos — the facts on this page are what protects it.
Common questions
Is my audio used to train AI models?
Not by us: we do not train models on your recordings or transcripts. The speech and language providers process them under their standard API terms.
Where is my data stored?
On Amazon Web Services in the Tokyo region. The transcription and language providers that process it are mainly in the United States.
How do I delete my data?
Delete a single transcript from its page or the library — its audio and video go with it — or delete the whole account from Profile. Both are immediate and you do not need to write to us.
Can someone else open my transcript?
No, unless you create a share link for it. Share links are read-only and can be revoked.
Is JustTranscribe HIPAA compliant or SOC 2 certified?
No. There is no HIPAA agreement and no SOC 2 or ISO 27001 report. Do not upload protected health information.
How long do you keep my files?
Video for 3 days. Audio while you use the account, removed after 7 days without a visit (kept while a Pro subscription is active). Transcripts until you delete them.
Who do I contact about security or privacy?
Write to hello@justtranscribe.ai. To report a vulnerability, use the same address and include the steps to reproduce it.