Azure AI Foundry HIPAA: Are Claude and DeepSeek Covered?

On this page

Azure's HIPAA BAA covers in-scope Azure services, but Microsoft does not publish a model-by-model list for Azure AI Foundry, and Claude in Foundry also falls under Anthropic's own BAA. On Concentrate, Claude 5.5, DeepSeek V4.1 Flash, GPT-6.1, Grok and Kimi all run on Azure as BAA-covered routes under one BAA, with every covered route listed publicly.

Does Azure sign a HIPAA BAA for AI Foundry?

Yes, for in-scope services. Microsoft says it "offers a HIPAA BAA as part of the Microsoft Product Terms" and that there is "no separate contract to sign" (Microsoft HIPAA offering). The open question for most teams is which Foundry models count as in scope.

Is every Foundry model covered by Microsoft's BAA?

Not clearly. In a July 2026 answer on Microsoft Q&A, Microsoft staff said Microsoft "does not publish a model-by-model list" and that a model "should not be assumed to be approved for regulated healthcare workloads" unless Microsoft documents it (Microsoft Q&A). That thread asked specifically about DeepSeek, Mistral and partner-hosted models, and it is one of several similar questions on the forum.

Is Claude in Azure AI Foundry covered?

Claude in Microsoft Foundry is covered by Anthropic's BAA, which your organization's Primary Owner signs; the BAA only covers the single organization that accepted it (Anthropic). So using Claude with PHI on Foundry directly means checking two agreements: Microsoft's and Anthropic's.