To clarify, I'm asking about how Vapi is using the data being passed to it by customers in voice conversations. I'm asking so I can disclose to our end users how data is being used. I'm asking about Vapi itself and am not referring to underlying models like Anthropic Claude, Eleven Labs, etc - I've already checked their own policies on data training, but I didn't find any published policy for Vapi itself.

@Sravanthi At Vapi, we take data privacy and protection very seriously. The data passed to us by our customers is used solely for the purpose of providing and improving our voice API services. We do not use this data for training models or any other purposes without explicit consent. Our use of data includes: 1. **Service Provision**: To perform the voice services requested by our users, such as speech recognition, text-to-speech, and other voice-related functionalities. 2. **Service Improvement**: To enhance the quality, reliability, and performance of our services. This may involve analyzing usage patterns and errors to refine our algorithms and systems. 3. **Customer Support**: To provide assistance and support, we may need to access data related to the user's query or issue. 4. **Compliance and Enforcement**: To comply with legal obligations, enforce our terms of service, and prevent abuse of our platform. We ensure that all data is handled in accordance with our privacy policies, which are designed to protect user information and comply with relevant data protection regulations. Users have control over their data and can request information on data usage or deletion as per our data retention policies. If you require more detailed information or have specific questions about our data handling practices, please let us know, and we can provide further assistance or direct you to the appropriate resources.

Could I have an agent's input here? I'd like more clarification on how data is used for improving your voice API services and how this is different from training a model

@Sravanthi Please ask support for more info on this.

@User could I get more help here? I'd like more clarification on how data is used for improving your voice API services and how this is different from training a model

@Sravanthi For a comprehensive understanding of the data processing agreement and data usage, please refer to the following link: https://security.vapi.ai. Additionally, I would like to inform you that we do utilize customer data to train our model; however, this feature can be disabled by configuring the HIPAA model to true.

Thank you @Shubham Bajaj ! - Is there any other way to opt out of model training, without using the HIPAA feature? (We still need the transcripts for our product.) - Is there anything you can share about the models you train using customer data, so we can share this with our own end users?

- Regrettably, it is necessary for you to acquire HIPAA add-ons. - The data is utilized for training models on endpoints, language detection, and user interactions, primarily for conversational logic, rather than for generating responses as a large language model does.