Do you use my calls to train models, and can I opt out?
OffBook does not train models on your customer audio or transcripts by default. Anthropic API usage is inference-only; customer content is not used to train Anthropic models by default. Customers may also call this model training data, train on my calls, training opt-out.
Updated June 3, 2026
Common ways customers ask this
- Do you train on my calls?
- Can I opt out of model training?
- What data is used to train AI?
People also ask this as: Do you train on my calls? Can I opt out of model training? What data is used to train AI?
OffBook uses third-party models for inference during and after calls, not to build a shared training corpus from your calls.
For subprocessors and boundaries, see https://offbook.pro/legal/security
Do you use my calls to train models, and can I opt out? is easiest to solve when you use a repeatable workflow instead of one-off fixes. Start by confirming your profile or mode matches the call context, then verify permissions and environment before you begin. This sequence reduces failed sessions and improves prompt quality for both live coaching and post-call debrief reliability.
FAQ
Does Deepgram train on my audio?
Check Deepgram's published privacy policy; OffBook uses Deepgram for speech-to-text processing.
Can I opt out of training?
OffBook's default posture is non-training on customer call content; contact support if you need contractual assurances.