Do not sell or share my personal information · Privacy choices, download or delete my data
We collect minimal data necessary to operate the Service. This includes your email address (via Whop at purchase), your licence key and subscription status, and usage data such as credit consumption.
Your licence key is stored in your browser (localStorage) so you stay logged in, and is sent to our servers with each request to verify your subscription and enforce credit limits. We send your licence key email via Resend, our transactional email provider.
If you use GPU transcription or speaker recognition, we also process the audio of the videos you choose. Section 5 explains how.
If you claim the free video, we also check your email, browser and network so each person gets it once. Section 16 explains what we keep.
We use your data to provide and improve the Service, process payments, and communicate with you about your subscription.
Only if you opt in, we also record anonymous facts about the videos you make (section 13) and may include them in trend reports we sell (section 14). Both choices are off unless you turn them on, and the Service works the same either way.
All payment processing is handled by Whop. We never see or store your credit card details. Please refer to Whop's privacy policy for information on how they handle payment data.
The Service uses the Anthropic Claude API (via our servers) to process your requests, including Supercut Builder (section 8). Your prompts are sent to Anthropic's servers. Please refer to Anthropic's privacy policy for details.
If you use the YouTube channel fetch or transcript features, requests are made to the YouTube Data API via our server proxy. No YouTube data is stored by us on our servers; transcripts you save are stored locally in your browser.
If you use GPU transcription or speaker recognition, the audio of your videos is stored for a short time in private Cloudflare R2 storage and processed on GPU servers rented from Vast.ai (section 5).
You can ask the app to transcribe your own videos and to recognise who is speaking in them. Your browser first takes the audio track out of the video and turns it into a compressed audio file. Only that audio file is uploaded. The video picture never leaves your device.
The audio goes from your browser straight to private Cloudflare R2 storage. The storage is not public, and the links used to upload and download files expire.
A GPU server rented from Vast.ai then processes the audio. We only rent machines that Vast.ai lists as verified. The machine transcribes the audio with Whisper and, if you asked for speaker recognition, tells the speakers apart with pyannote. Both models run on the rented machine itself, so your audio is not sent to any other AI service. The machine deletes its working copy when the job ends and is destroyed once it has no more work. Vast.ai hosts are independent providers and may be in a different country from you.
We use your audio only to produce the transcript and speaker labels you asked for. The result contains the transcript with word timings. For speaker recognition it also contains when each speaker talks, a voiceprint for each speaker (section 6) and a sample clip of up to 5 seconds per speaker, so you can hear who is who.
Your browser downloads the result and saves it on your device. Once it has, the audio and the result are deleted from storage. If that does not happen, for example because you do not open the app again, the audio is deleted automatically within 1 day and the result within 7 days. Cancelling a job before it starts deletes the uploaded audio straight away.
To run the queue and to charge or refund credits, our servers keep a short record of each job for 7 days after it ends: the file name, size and length, the type of job, its status and the credits charged. We also count the audio minutes you use, for the daily and trial limits, for up to 8 days. These records contain no audio, transcript text or voiceprints.
A voiceprint (speaker embedding) is a list of numbers calculated from a person's voice. The app uses it to recognise the same voice in other episodes and name that speaker automatically. Voiceprints are biometric data. Privacy laws, including the Australian Privacy Act and the GDPR, treat biometric data used to identify people as sensitive.
Voiceprints are created only when you ask for speaker recognition on a file. Transcription on its own does not create them.
Your voice library (the names you give speakers, their voiceprints and their sample clips) is stored only in your browser on your device, together with the speaker data of each transcript. We never store voiceprints on our servers. The only copy that passes through our systems is inside the temporary result file described in section 5, which is deleted as described there. Voiceprints are never sent to Anthropic or any other AI service.
Matching happens in your browser. When a known voice is recognised with high confidence, the app adds the new episode's voiceprint to that voice so later matches are more accurate. If you export a backup of your library, the file contains voiceprints, so keep it private.
Only upload recordings you have the right to process. If other people can be heard in them, you are responsible for having their consent where the law requires it, especially before you name them or keep their voiceprints in your voice library. For these recordings you decide what is processed, and we process it on your behalf.
If someone asks you to stop recognising their voice, delete their voice and the transcripts that contain it (section 9). We do not keep voiceprints, so a person heard in a recording should ask the customer who processed it to delete them.
Cutting clips and Auto Cut run entirely in your browser, on your own computer. Your video is not uploaded for them. So that the app can find a video again, it may remember the file's name, size and length and, in browsers that allow it, permission to reopen the file. This also stays in your browser.
Supercut Builder uses the Anthropic Claude API through our servers, the same way our other AI features do. It sends your brief and the text of the transcripts it works on to Anthropic, together with the names of the speakers you named in those transcripts, so it can search by who is talking. It does not send audio, video or voiceprints. When you have named speakers, the transcript text sent by any AI feature includes those names.
Your transcripts, word timings, voice library, list of GPU jobs and settings are stored in your browser (localStorage and IndexedDB) on the device you use. They are not copied to our servers or to your other devices, and we cannot see or delete them. You can delete them at any time:
We cannot recover data you delete from your browser.
We use localStorage and IndexedDB in your browser to store your licence status, settings and the library data described in section 9. We do not use tracking cookies or third-party analytics. The optional usage data in section 13 is collected by our own servers, only if you opt in, and is not linked to your account.
We retain your account data for as long as your subscription is active. You can delete your data yourself at any time (section 12) or by contacting us.
Your privacy choices and the record of when you made or changed them are kept while you have an account, so we can show what you agreed to, and deleted when you delete your data. Anonymous usage facts (section 13) are kept for up to 13 months.
For GPU jobs, uploaded audio is deleted once your browser has downloaded the result, and always within 1 day. Results are deleted once downloaded, and always within 7 days. Job records are deleted 7 days after the job ends. Data in your browser stays there until you delete it.
Under Australian Privacy Act and GDPR (where applicable), you have the right to access, correct, or delete your personal data, and to withdraw a consent you gave. Depending on where you live (for example California) you also have the right to opt out of the sale or sharing of your personal information.
You can do this yourself: press Privacy at the top of the app to change your choices, download everything our servers hold about you, or delete it (optionally closing your account). You can also contact us. We will not treat you differently for using these rights.
Data kept only in your browser, including your transcripts, voice library and voiceprints, is under your control and we cannot access it. Use the steps in section 9 to delete it.
If you tick "Help improve ViralDNA", our servers record anonymous facts each time you use Autopilot: whether you made a long video or a short, the length you asked for, the title of the theme (for example "Wildest Drug Stories"), how many clips and episodes were used, how long the finished video is, and whether it was finished, downloaded or exported. The time is rounded to the hour.
These facts are stored without your account, licence key, email or IP address, so they cannot be linked back to you. We never record your brief, transcripts, audio, video, voiceprints, speaker names or channel. Theme titles that look like contact details or links are thrown away.
We use these facts to understand what creators are making and to improve the Service. You can turn this off at any time under Privacy; collection stops straight away.
If you also tick "Include them in trend reports we sell", your anonymous usage facts may be counted in reports we sell to others, for example businesses that want to know which video themes creators make most. Reports only contain totals across many people. A theme that fewer than 5 people made is left out, so a rare theme cannot point to one channel.
We do not sell your name, email, account details, licence key, transcripts, videos, audio, voiceprints or anything that identifies you or your channel. This choice needs the first one, and it is off unless you turn it on.
Use the Do not sell or share link (or the Privacy button in the app) to turn off "Include them in trend reports we sell" at any time. From then on nothing you make is counted in reports we sell. Reports already sold cannot be recalled, but they only ever contained anonymous totals.
You can make one free Autopilot video without a card. To stop the same person claiming it again and again, we check a few things when you ask for it:
We store the email address on your free account, like a paid account, and the country your network was in. The device id, fingerprint and IP address are only stored as keyed hashes (scrambled with a secret key), never in readable form. The hashes of your email, device id and fingerprint are kept so the free video cannot be claimed twice; the count for your network is deleted after 30 days, and an unused code after 15 minutes.
The free video ends with a 3-second "Made with ViralDNA" card. Paid plans have no end card.
For privacy questions, contact us at alxy369@gmail.com.