WATGAN — anonymous, untraceable creative social.
All video is GAN-filtered on your device. All voice is masked on your device.
The server sees nothing. Stores nothing. Logs nothing.
No single point of failure. Relay nodes help peers find each other.
If one dies, the network continues. Anyone can run a relay.
Your identity is a cryptographic keypair encrypted with your passphrase.
It never leaves your device. No email, no phone, no server accounts.
preview
loading transcription model...
welcome to watgan
Create your anonymous identity. No email. No phone. Just a handle and a passphrase.
Your encrypted keypair is stored on the P2P network — sign in from any device with your handle + passphrase.
welcome back
Signing in as
your privacy tiers
WATGAN has three privacy tiers. You control which level you operate at.
tier 1 — local only (default)
Create filtered videos and save to your device. Nothing is uploaded. Nothing leaves your phone.
Video filters destroy facial data. Voice filters destroy voiceprint. The raw camera/mic feed
is never recorded — only the filtered output exists.
tier 2 — ephemeral post
Post to the P2P feed with a decay timer (1hr–30 days). Your content is shared with peers
but automatically expires. Metadata passes through relay nodes that store nothing and log nothing.
Your identity is a cryptographic keypair — no email, no phone, no name.
tier 3 — permanent gallery
Choose to keep posts permanently as a public creative portfolio. Your handle and art are visible
at your profile URL. Media is stored on a server. This is the only tier where data persists
beyond your device. You can delete any post at any time.
Key principle: Every tier is opt-in. You start at Tier 1 by default.
No data is ever uploaded without your explicit choice.
transcription
Transcription uses Whisper AI running on our server to convert speech to text with word-level timestamps.
what happens to your audio
Your recorded audio (already voice-filtered if you have a filter active) is sent encrypted over HTTPS to our server. Whisper processes it and returns the transcript. The audio is deleted immediately — it is never stored, logged, or retained.
why server-side?
Browser-based transcription (WASM) is too slow on most devices. Server-side Whisper completes in seconds instead of minutes. If you have a voice filter active, the server only ever hears the masked version — your real voice never leaves your device.
You can cancel at any time. This choice is remembered for this session.
WATGAN