Azure OpenAI Assistants: What privacy features are true for the assistant API

Question

Hey
Reading the privacy policy we find what seems like a contradiction or a state to which we don't find a direct reference

On the one hand:
Reading this API documentation:

https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/assistant#supported-file-types

We note that to run an assistant a thread_id should be created, this is also required for acquiring an old conversation history without resending past exchanges with the assistant.

But this also means that Azure stores our conversation history, otherwise, it would not be able to continue those with only a thread_id. meaning the assistant API is not stateless.

On the other hand:
In this link

https://learn.microsoft.com/en-us/legal/cognitive-services/openai/data-privacy?context=%2Fazure%2Fai-services%2Fopenai%2Fcontext%2Fcontext
it says:
"Your prompts (inputs) and completions (outputs), your embeddings, and your training data:
*are NOT used for automatically improving Azure OpenAI models for your use in your resource (The models are stateless, unless you explicitly fine-tune models with your training data)."

Contradiction

We can agree that an assistant thread has a state, does azure provide data secrecy, and privacy, to assistants as it does to other chat completion mechanisms?

This is something our customers are not comfortable with, so we need a clear concise reference that would clarify what privacy features are true for the assistant API

Thanks

Answer 1

Hello @Evyatar Or

Thanks for reaching out to us. First, all features under Azure OpenAI will follow the same data and privacy policy.

Your prompts (inputs) and completions (outputs), your embeddings, and your training data:

• are NOT available to other customers.
• are NOT available to OpenAI.
• are NOT used to improve OpenAI models.
• are NOT used to improve any Microsoft or 3rd party products or services.
• are NOT used for automatically improving Azure OpenAI models for your use in your resource (The models are stateless, unless you explicitly fine-tune models with your training data).
• Your fine-tuned Azure OpenAI models are available exclusively for your use.

The Azure OpenAI Service is fully controlled by Microsoft; Microsoft hosts the OpenAI models in Microsoft’s Azure environment and the Service does NOT interact with any services operated by OpenAI (e.g. ChatGPT, or the OpenAI API).

Azure OpenAI Assistant uses thread_id to manage conversation state temporarily but does not imply permanent storage of conversation history for your usage. It's only temporarily to maintain the state of ongoing conversations. This context storage is essential for enabling coherent and relevant interactions during the conversation. And the data can be only access by your and who other has the valid subscription ID/roles.

https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/assistants#assistants-data-access

Conversation history, when stored, is generally retained only for the duration necessary to support the ongoing conversation and is not kept beyond that period unless explicitly required by user actions or configurations.

I hope this helps.

Regards,

Yutong

-Please kindly accept the answer if you feel helpful to support the community, thanks a lot.