How to Access Anthropomorphic Models in Azure Foundry Marketplace

Select Language:

If you’re using Anthropic Claude models through Azure AI Foundry and are concerned about data privacy, here’s what you need to know and how to manage your data effectively.

First, it’s important to understand that by default, prompts and outputs you generate with Anthropic Claude models are not used to train the model itself. Anthropic has made it clear that they do not utilize customer prompts or completions from commercial or API usage, including Foundry, for training their models. Training would only happen if you explicitly opt in for that purpose.

When it comes to data handling, Anthropic operates as a data processor. This means they manage and process your prompts and responses on their infrastructure. However, they might retain your data temporarily for specific reasons like safety checks, abuse detection, or ensuring the app runs smoothly. Usually, this retention lasts around a week to a month, but if your data is flagged for policy or abuse reasons, it could be stored longer.

It’s crucial to note that Azure’s “zero data retention” guarantee does not automatically apply to models provided by Anthropic. Microsoft provides the platform, handles deployment, and manages billing, but does not have control over how Anthropic processes or retains your data. Ultimately, the policies governing data processing and storage are dictated by Anthropic’s own practices and agreements, including their Data Processing Addendum.

Because these models are considered third-party SaaS offerings within Foundry, it’s wise to review Anthropic’s specific policies on data privacy and retention. For more details, check out the Azure documentation related to partner and community models, especially the section about Anthropic models, which offers more insight into data handling practices.

If your main concern is safeguarding sensitive information, the key is to understand that while your data isn’t used for training in most cases, it may still be processed and stored temporarily on Anthropic’s side. Always refer to the official policies to see how your data is managed, and consider whether these align with your privacy requirements.

Feel free to reach out if you have any further questions or need additional clarification. Taking these steps can help you manage your data responsibly while using advanced AI tools.