Hi,
Great questions - happy to walk through each one.
1. Custom S3 + Vapi storage
When you connect your own S3 bucket, call recordings (audio files) are redirected there instead of Vapi's servers. However, other data - transcripts, messages, summaries, and call metadata — still gets stored in Vapi's database.
If you want Vapi to store nothing on our side, you'd need to enable Zero Data Retention (ZDR) mode alongside custom S3. With both enabled, recordings go to your S3 bucket and all other call data is scrubbed from Vapi's systems - you still receive everything via end-of-call webhooks.
2. Requesting data deletion
Yes, a few options:
- Email support@vapi.ai with a deletion request - data not required for legal/business purposes is deleted within 30 days
- Use the DELETE /calls API to remove individual call records:
https://docs.vapi.ai/api-reference/calls/delete
- Enterprise customers also have access to bulk deletion
We're GDPR and CCPA compliant, so data subject rights requests are fully supported. For reference, default retention on pay-as-you-go is 14 days for call data and 30 days for chat data.
3. TTS/STT provider training policy
Vapi does not use your call data to train any AI models - no opt-out needed, this is the default.
For TTS/STT providers in our dropdown (Deepgram, ElevenLabs, etc.), we configure them all in no-training, zero-retention modes by default and opt out of any human review programs they offer. Audio is processed for the conversion and then discarded - it's not retained or used to improve their models.
For more detail on our security posture, SOC 2, and DPA templates, check our trust center at
https://security.vapi.ai.
Let me know if you have any follow-up questions!
Best,
Vapi Support