Is my data used to train AI models?
Not by ModelRiver. We don't use your prompts or responses to train AI models, and we don't sell or share your data. It's used only to run, secure and support the service.
What about OpenAI, Anthropic and the other providers?#
Your requests go to the providers you choose, using your own provider API keys. That means your existing agreement with each provider applies, including its data and training terms. ModelRiver doesn't change those terms. Check each provider's policy, and its settings for business or API accounts, if training matters to you.
Providers you connect aren't ModelRiver subprocessors; they're your own vendors.
Is anything sent to a provider on ModelRiver's account?#
One thing. With guardrails on (the default), a prompt that ModelRiver's own checks can't classify is sent to OpenAI's moderation service on ModelRiver's account, only to check it for safety. OpenAI is listed as a subprocessor for this purpose. Prompts blocked by guardrails are never stored.
Does ModelRiver use my data in any other way?#
Our terms allow us to use aggregated, anonymized data to improve the service. That's the only use beyond running, securing and supporting your account.
How can I limit what's stored at all?#
Each project can turn off saving full request and response text. See What data does ModelRiver store?