How do I use haystack for private data?

[sohanasarah]

I want to use haystack pipelines to train the models with private data. As I am using haystack locally, I assume it is safe. My question is, how can I ensure that data is not stored somewhere while processing and training with confidential data?

[mayankjobanputra]

Haystack doesn't store/stream any of your data anywhere. At best, we collect the events, nothing else. If you even want to disable the events, also you can do it by export HAYSTACK_TELEMETRY_ENABLED=False

[sohanasarah]

Thank you for the clarification.

[julian-risch]

Confirming what Mayank said, with Haystack your data always stays on your local machine unless you explicitly use an external API. For example, for your use case you would probably choose a document store other than PineconeDocumentStore and you would not use the PromptNode with an OpenAI API key but only with local models. In short, as long as you do not explicitly provide API keys to Haystack, your data will stay on your machine.
Telemetry does not send any private data but just anonymized usage statistics so that we can learn how many users run a Haystack tutorial for example. Our documentation page lists details about telemetry here: https://docs.haystack.deepset.ai/docs/telemetry

[sohanasarah]

Thank you very much. This answers my question.