Privacy

Your voice never leaves your machine. Speech recognition and AI cleanup run as local models on your Mac — there is no Hushkey server that ever sees your audio or the text you dictate.

That claim is easy to make and easy to check, so here is the complete list.

Everything the app sends over the network

• Model downloads. On first run (and when you pick a new model), Hushkey downloads the speech and cleanup models from Hugging Face or a configured mirror — models come down, nothing about you goes up.
• License checks. When you activate, validate, or deactivate a license, the app calls Lemon Squeezy's license API with your license key and a device identifier — never audio or text.
• Crash reports. Only if you opt in after a crash, one report per incident: a technical crash trace and system info — never your dictated text or audio. Nothing is sent without your yes.
• This website. hushkey.ai uses Cloudflare Web Analytics: no cookies, no cross-site tracking, aggregate visit counts only.

What the app never does

• No telemetry or usage analytics in the app — not even opt-in.
• No accounts. There is nothing to sign up for and no profile of you anywhere.
• No audio or transcript upload, ever. Dictation works with Wi-Fi off.

Questions about any of this? Email [email protected] — a human reads it.