privacy policy.

Audio you stream to our API is held in memory only for as long as it takes to transcribe it, and is discarded when the connection closes. It is never written to disk and never stored.

Transcripts are returned to you over the same connection and are not saved to our database. We cannot show you the contents of a past call, because we do not have them.

We do not use your audio or transcripts to train, fine-tune or evaluate models, and we do not review them.

To bill accurately and help you debug, we record one row per call containing no speech content:

• a call identifier, and the project and API key the call belonged to
• seconds of audio processed and seconds the connection was open
• start and end time, the model used, and whether the call completed, failed or was rejected
• an error code when a call fails — never a transcript

These records are what you see on your usage dashboard. We keep them while your account is active and for as long as we need them for billing, tax and accounting records.

Speech recognition runs on our own GPU servers in Amazon Web Services (Oregon, us-west-2). The speech model is self-hosted — your audio is not sent to OpenAI, Anthropic, Google, or any other model provider for transcription.

If you connect from outside the United States, your audio is processed in the United States.

The comparison playground and the "try it" demo on our home page deliberately send the same microphone audio to a third-party provider (currently Deepgram) at the same time, so you can compare results side by side. That audio is handled under that provider's own privacy terms, not ours.

This only happens in the browser demo and the console playground. Calls to our API are never forwarded to another provider.

Signing in with Google gives us your name, email address and profile picture. We store your project names, team members, and the invitations you send.

Payments are processed by Stripe. Card details go directly to Stripe and never reach our servers; we keep a Stripe customer reference and a record of credits added to your account.

API keys are shown once at creation and are not recoverable — we store only an identifier and a short display prefix, never the key itself.

We set a session cookie so you stay signed in. We use Microsoft Clarity to understand how the site and console are used — it sets analytics cookies and captures usage data such as pages visited, clicks, and scrolling. API keys are masked and never appear in these recordings. We do not use advertising cookies, and we do not sell or share data with advertisers.

For audio you send us, you are the data controller and we act on your instructions. You are responsible for having a lawful basis to record and transcribe the people you capture, including any notice or consent your jurisdiction requires.

Because we do not retain audio or transcripts, requests to delete the contents of a particular call are satisfied automatically — there is nothing to delete.

Audio and transcripts: not retained. Usage records, projects and account data: deleted within 30 days of account closure, with backups purged within 60 days, except records we must keep for tax and accounting purposes.

Waitlist signups are kept until we launch or you ask us to remove you, whichever is first.

Traffic to our API and console is encrypted in transit. API keys are signed tokens that carry no secret material about other customers, and can be revoked instantly from your console.

We will announce material changes to this policy by email or in the console before they take effect.

Questions, or a data request: privacy@autoloops.ai.