Is my business data used to train AI models?
It depends on the vendor, and there are two questions rather than one: whether the product you signed up to trains on your content, and whether the model provider behind it does. A vendor can answer no to the first while the second is still true.
How it actually works
Most business-facing AI products now state that customer content is not used to train shared models, and most major model providers offer no-training terms for API traffic — which is a different arrangement from their consumer chat products, where the default often runs the other way.
The question worth asking is how isolation is enforced rather than whether it is promised. Filtering results by account after retrieval, or instructing a model not to reveal other customers' data, are both weaker than filtering inside the database query itself, where there is no code path that could return another account's content in the first place. RegenticFlow does the last of these.
Exit terms are the other half. Whether you can export everything in a machine-readable format, and whether deletion actually cascades to documents, conversations, assistants and leads, tells you more about how your data is treated than any paragraph in a privacy policy.
What it can't do
- “Your data never leaves” is not a claim anyone using a hosted model can honestly make, and we do not make it. Your text is processed by OpenAI under a no-training arrangement.
- We hold no security certifications — no SOC 2, no ISO. Where that matters, it matters, and it is stated on our security page rather than omitted.
- Retention of public conversation transcripts is configurable between 7 and 365 days, defaulting to 90.
Related
See it on your own documents
One public assistant page and 100 customer messages a month, free, with no card required.