Security & Trust

Transcripts are sensitive: customer calls, voice memos, interviews, lectures. Here's exactly what happens to your audio when it passes through Whipscribe — short enough to read in a meeting, specific enough to forward to your security team.

At a glance

Data handling

What we receive

• The audio/video you upload or the URL you paste.
• The transcript we produce from it.
• Job metadata: duration, language detected, status, timestamps, your email (if signed in).

Where it lives

Everything is hosted on our own servers at Vultr (EU + US regions). Speech-to-text runs on the same host family — no third-party AI vendor ever sees your audio. The transcription engine is self-hosted. The storage layer is either Vultr Object Storage (S3-compatible) or an encrypted disk volume, depending on tier.

We do not use OpenAI, Anthropic, Google Cloud Speech, AWS Transcribe, or any other hosted model for the transcription itself. When a user opts to forward a finished transcript to an LLM (e.g. "send to Claude"), that action is their own — we don't proxy it and we don't store the result.

Retention

Ask us to delete sooner at contact@neugence.ai — we honour deletion requests within 7 days.

Your data is never used to train AI models

Not our models, not anyone else's. Uploads, transcripts, and derived artefacts are used solely to deliver the service you asked for. We do not sell data, and we do not share it with advertisers or analytics brokers beyond aggregate product metrics.

Authorization

We treat Whipscribe as a tool, not a library. You upload content, we transcribe it. That model depends on three things:

• Rights attestation on every upload. The form's "I own the rights to this recording, or have permission to transcribe it" checkbox is a required affirmative consent. Submissions are blocked without it.
• URL paste is limited to Creative Commons / public-domain sources. We will not fetch copyrighted YouTube, Spotify, or paywalled content on your behalf. The inline note on the paste form says so; violating it is a ToS breach on your side.
• No re-hosting. Your transcripts live in your vault, not in a shared searchable library. There is no public transcript index or aggregated corpus of other people's audio.

Consent & jurisdiction

If you're recording a call, a lecture, or an event with other people in it, the law treats you as the party doing the recording. Whipscribe is the tool; you are responsible for obtaining consent from the other parties involved.

Twelve US states require all-party consent for recording audio (California, Connecticut, Florida, Illinois, Maryland, Massachusetts, Michigan, Montana, Nevada, New Hampshire, Pennsylvania, Washington). The EU's GDPR requires a lawful basis plus notice. When in doubt, disclose that the call is being transcribed.

Coming soon: auto-generated consent disclosure you can paste into the chat of a meeting or live Space. See the roadmap below.

Infrastructure

• Hosting: Vultr Cloud Compute, EU (Frankfurt) and US (New Jersey) regions.
• Network: app servers reachable only through Tailscale; public traffic terminates at Nginx + Caddy with TLS 1.2+ (HSTS on).
• Auth: Firebase Authentication for user accounts, signed tokens for short-lived anonymous / guest sessions.
• Payments: Razorpay. We never see or store full card numbers; the PCI scope lives with Razorpay.
• Backups: encrypted snapshots of metadata; never of raw audio (which is ephemeral by policy).
• Engine: self-hosted speech-to-text with the model weights on our own hosts. No inference goes to third-party cloud AI APIs.

Sub-processors

Compliance roadmap

Responsible disclosure

Found a security issue? Please send it to security@neugence.ai (GPG on request). We commit to:

• Acknowledge within 2 business days.
• Provide a triage / ETA within 7 days.
• Credit reporters publicly once the fix is live (opt-out available).
• Not pursue legal action against good-faith research that respects user privacy and doesn't destroy data.

Whipscribe is built by Neugence Technology Pvt. Ltd. See also the Privacy Policy and Terms of Service.