Azure OpenAI GPT-5.6 Sol Data Zone EUR — HDS scope and processing locations

We are a French digital health company preparing an Azure OpenAI deployment that may ultimately process personal health data subject to French HDS requirements.

Our Azure OpenAI resource is located in France Central. The deployment uses:

• Model: gpt-5.6-sol
• Model version: 2026-07-09 (GA)
• Deployment type: Data Zone Standard
• Data zone: EUR

We have reviewed Microsoft’s current HDS Certificate No. 1042480-3 and the HDS Representation of Guarantees (July/August 2026) available through the Microsoft Service Trust Portal.

Before processing any real patient data, we need written confirmation from Microsoft of the following:

1. Can gpt-5.6-sol with Data Zone Standard EUR process prompts or completions in Switzerland North, or in any other location outside the European Economic Area (EEA)?
2. What are all possible processing/inference locations for this specific deployment configuration?
3. Are all possible processing locations for this deployment covered by Microsoft’s current HDS certification applicable to Azure?
4. Does HDS Certificate No. 1042480-3 apply to this specific Azure OpenAI / gpt-5.6-sol / Data Zone Standard EUR architecture?
5. What contractual HDS terms apply to a French Azure customer processing personal health data? Is any additional HDS agreement, addendum, enrollment, customer commitment, or specific configuration required before processing personal health data?
6. Microsoft’s HDS Representation of Guarantees states that customers subject to HDS are required to store their data in the EEA. Since Switzerland is not part of the EEA and Switzerland North does not appear among the datacenters listed in HDS Certificate No. 1042480-3, please specifically confirm whether this deployment can process Customer Data there.
7. What is the current retention period for prompts and completions under standard Azure OpenAI abuse monitoring?

We have separately submitted an application for Modified Abuse Monitoring. We understand that, if approved and correctly applied, this should prevent abuse-monitoring storage and human review of prompts/completions. We intend to verify activation through the applicable resource capability, including ContentLogging:false where applicable.

Because these questions require confirmation of Microsoft’s internal processing locations and the exact contractual/certification scope, the public documentation available to us does not appear sufficient to establish the answer.

Please escalate this request to the appropriate Azure OpenAI privacy/compliance and HDS team and create a private support case if required.

We can provide our Azure subscription and resource identifiers privately if required. We will not post them publicly.

No real patient data is currently being processed through this deployment. All Azure OpenAI qualification testing remains synthetic pending resolution of these questions.