Privacy

You don't need an account to use TranscriptFox, and we don't use tracking cookies or advertising networks.

Guests and Free accounts may see one labelled advertisement above a finished website transcript, which may include links to Amazon and eBay. These are plain links: nothing loads from the advertisers, and nothing about you, your link or your transcript is sent to them. If you follow one, the advertiser’s site may set its own cookies, and we may earn a commission if you buy something there. As an Amazon Associate, TranscriptFox earns from qualifying purchases. Paid plans, trials and transcript passes never show advertisements.

You can create an account with a passkey, using your device’s screen lock. We store a random account identifier and the passkey’s public key. Your fingerprint, face data, device PIN and private key stay with your device or passkey provider. You can add another passkey while signed in. We cannot recover a passkey account if every passkey is lost, so keep one in a password manager or add one on another device.

When email sign-in is offered on the account page, it sends a single-use link through Resend, our email delivery provider. We share your email address and sign-in message with Resend to deliver it. Links expire after 15 minutes and must be confirmed in the browser that requested them. We keep your verified address to recognize your account; pending addresses and hashed links are removed after expiry during later account requests. We do not use these addresses for marketing.

When you sign in with Google, Microsoft, Yahoo or, when available, ChatGPT, that service shares a stable account identifier, your name, your email address and, when it has one, a link to your profile picture with TranscriptFox. We keep these to recognize your account and show who is signed in. We copy a small version of your profile picture when you sign in, so our pages never load it from that service. We never join two accounts just because they share an email address. We use secure sign-in cookies: passkey attempts expire after five minutes, Google, Microsoft, Yahoo and ChatGPT sign-in attempts after ten minutes and the signed-in session after eight hours. Signing out removes the current session. A sign-in in the Android app lasts until you sign out there or it goes unused for 180 days. Expired sign-in records are removed during later account requests. We do not keep their sign-in tokens, and we do not receive your emails, files, contacts, ChatGPT conversations or passwords. Free accounts do not save your transcripts.

Paid plans and one-time transcript passes are billed by Stripe, our payment provider. Stripe collects your card or other payment details directly; we never see or store your full card number. We keep a Stripe customer, subscription or payment reference with your account, plus your plan, pass balance, status and renewal date, so we can give you the right allowance. Stripe sends us payment updates. Stripe’s own privacy policy covers the payment details it holds.

On Pro, every transcript you get on the website or in the app, including transcripts of uploaded files, is saved to your account so you can open it again from your Transcripts page. On Plus, you can manually save transcripts to a 30-day history on your account. Saved transcripts are private by default so only you can see them. If you choose to create an unlisted share link for a saved transcript, anyone with that link can read it until you turn sharing off or delete the transcript; uploaded files are never made publicly indexable. On Pro, we keep your saved transcripts until you delete them, including after your plan ends; deleting one removes it from your account right away, together with the audio copy of an uploaded file. Each saved transcript is stored as its own copy with your account, so it does not change later. Transcripts you get through the API or MCP are not saved. We never keep the uploaded file itself, only the compressed audio copy for the player described below, and nothing of an uploaded PDF or image.

TranscriptFox no longer offers AI summaries or Ask AI. Answers made before they were removed stay with the account or public link they belong to only until their original expiry, up to 180 days for a summary of a public link and up to 30 days for everything else, and are deleted sooner when you delete that saved transcript or your account. Nothing new is sent to a language model for a summary, and we do not use your transcripts or questions to train any model.

When you paste a link, our server fetches the captions for that video or audio. If there are none, or if the captions are incomplete or uncertain, it downloads the audio and sends it to speech recognition services to turn the speech into text. By default across Free, $1 Pass, Plus and Pro, speech recognition uses Microsoft’s MAI-Transcribe-2 model through OpenRouter. When Ukrainian or Russian speech is detected, our server may also check that audio with ElevenLabs’s Scribe v2 model through OpenRouter and use its transcript when the spoken language is Belarusian. Groq and Deepgram handle Hindi and Urdu writing-system checks, music and unclear-part verification, and fallback speech recognition when the primary speech service is unavailable (or Deepgram when Groq is unavailable and emergency fallback is enabled). The source download is deleted after processing. We keep a copy of each link transcript so the next request for the same video is instant. When a pasted link fails, we keep that link, its site, the step where it stopped and the error for 30 days, so we can find out why and fix it. This record is not tied to your account or IP address, and uploaded files and app recordings are never kept this way.

If you upload a file, we send its audio to the same speech recognition services. The uploaded file is deleted after processing, whether processing succeeds or fails. Its transcript is held temporarily so your browser can retrieve it, and is not added to the shared transcript cache.

If you upload a PDF or an image to transcribe it, it is read only on TranscriptFox’s own servers. Documents and their text are never sent to Groq, Deepgram, OpenRouter, Microsoft, ElevenLabs or any other outside service, and are not used to train any model. Pages with handwriting are sent as images over our private network to another of our own servers, which reads them and keeps nothing. The file and the page images made from it are deleted after processing, whether it succeeds or fails. The text is held temporarily so your browser can retrieve it, is not added to the shared transcript cache, and is saved to your account only as described above. Show original displays the file from your own device and does not upload it again. Our logs for documents record file types, page counts and processing steps, not file names, titles or text.

OpenRouter passes audio requests to Microsoft Azure for MAI-Transcribe-2 and to ElevenLabs for Scribe v2. Every OpenRouter speech request requires Zero Data Retention endpoints and denies data collection and model training. On Plus and Pro, speech recognition runs up to your plan’s monthly audio allowance.

The speech recognition services receive the audio in parts of up to 20 minutes from our server. They do not receive your file name, account or IP address. OpenRouter Zero Data Retention endpoints for Microsoft Azure and ElevenLabs do not retain audio or transcripts after processing or use them to train models. We ask Deepgram not to keep the audio for improving its models. Groq says it does not keep request data by default, except logs kept for up to 30 days when it checks reliability or abuse. We do not use your files, recordings or transcripts to train any model.

The Android app can record your voice with the microphone, only while you have started a recording. A recording is sent to us like an uploaded file and handled the same way. The app keeps it on your phone so you can retry or listen back, until you discard it, delete its transcript or clear your transcripts. When you delete the transcript of a recording or uploaded file in the app, the app also asks us to delete our audio copy of it.

On Pro and dearer plans, you can upload interview recordings on the Interviews page. They are transcribed only on TranscriptFox’s own servers; interview recordings and their transcripts are never sent to Groq, Deepgram, OpenRouter, Microsoft, ElevenLabs or any other outside service, and are not used to train any model. We keep the recording, a playback copy, the transcript and the history of your changes until you delete the interview or your account, also after your plan ends. Interviews are kept in encrypted backups for up to about three months, so a deleted interview can stay in those backups until they expire. Our logs for interviews record sizes, lengths and processing steps, not file names, titles or words. Your account data download includes your interview transcripts and change history.

If your account belongs to a TranscriptFox Government organization, you can make draft meeting minutes from a saved transcript of an open public meeting. To write the draft, our server sends the transcript and the text read from the agenda and documents you add to a language model through Groq, or through OpenRouter or OpenAI when Groq is unavailable. Documents you link are downloaded by our server and documents you upload are read on our servers; only their text is kept, and the files are deleted after reading. Do not use it for closed meetings or confidential documents. We keep each draft, its copy of the transcript, the document text and the history of your changes until you delete the draft or your account. Deleted drafts can stay in our backups for up to about three months. Our logs for minutes record counts, sizes and processing times, not titles, names, links or text. Your account data download includes your drafts and change history. We do not use your meetings or documents to train any model.

For YouTube links, the page shows YouTube’s player in its privacy-enhanced mode (youtube-nocookie.com). Your browser loads it from YouTube, so YouTube receives your IP address, and its own privacy policy applies to the video. We do not download YouTube audio for playback.

For the audio player, we keep a compressed audio copy. Audio from a link is kept for up to one hour and can be loaded again from the link. Audio of an uploaded file or an app recording is kept for 24 hours if you were not signed in when you uploaded it, 30 days on a Free account, and for as long as your Plus or Pro plan stays active. When a paid plan ends, its copies are kept for 30 more days. If storage fills, copies from people who were not signed in are removed first, then Free copies; copies on a paid plan are not removed for space. Each copy has a private, unguessable playback address. Anyone with that address can listen until the copy is deleted.

If you request a TranscriptFox Government demo, we first email a confirmation link to the address you entered, through Resend, so we know the address is yours. A request that is not confirmed within 24 hours is deleted without reaching us. Once you confirm, we keep your name, agency, work email and message for up to one year so we can reply and follow up, and we email them to our own inbox through Resend. We don't add you to a mailing list or share them with anyone else.

Like most websites, our servers keep short-term technical logs, such as IP addresses and errors, to keep the service running and stop abuse.

Our public API and MCP server read source captions or previously saved transcripts; they do not start speech recognition or expose upload results. Free previews cover the first 10 minutes. The daily transcript allowance is shared with the website, per connection when anonymous and per account when signed in. Free accounts also share their connection’s allowance. We store usage buckets tied to an account or a hashed connection for up to two days, removed during later requests, to count successful transcripts and protect service capacity. Successful text responses also store a hashed recording key and delivery time for up to two days, removed during later transcript requests, so pages and retries do not spend extra transcript credits. Existing access logs may contain addresses. See API and MCP for details.

We use Peeklog to understand visits and improve the site, without tracking cookies. It keeps a random visit identifier in your browser’s session storage, which is cleared when you close the tab. It records page addresses, referrers, browser and device details, approximate location, and interactions such as page visits and link clicks. Page addresses include the media link you paste, because the transcript page shows it in its address. Transcript text and uploaded files are not sent to Peeklog. When you are signed in, we associate these visits with your account, and we keep that association for one year or until you delete your account.

You can download a copy of your account data or delete your account at any time on your account page. Deleting it removes your saved transcripts, audio copies, sign-in methods and plan, and cancels any subscription. For other privacy requests, email vlad@transcriptfox.com.