Overview
Azure Machine Learning (AML) provides a secure, compliant environment for building, training, and deploying machine‑learning models. This page outlines how AML handles customer data, the privacy safeguards in place, and your responsibilities as a data owner.
Data Collection
AML collects only the data you explicitly upload—datasets, model artifacts, experiment logs, and configuration files. System‑generated telemetry (e.g., service health, usage metrics) is also collected in compliance with the Microsoft privacy policy.
Telemetry is aggregated and stored in a separate, non‑identifiable store. It helps improve service reliability and performance but is never used to infer personal data about your organization.
Data Usage
All data processing occurs within the Azure region you select. AML does not copy or move your data outside that region unless you explicitly configure a cross‑region workflow.
When you deploy a model, AML creates a secure endpoint that only serves the model output. Input data is not stored permanently unless you enable logging.
Data Retention
Datasets, experiment artifacts, and models remain in your storage account until you delete them. AML does not impose additional retention periods.
When you delete resources via the Azure portal or CLI, AML triggers a secure purge of related metadata. Physical storage is cleared according to Azure’s storage lifecycle policies.
Customer Data Ownership
You retain full ownership and control of all data, models, and results stored in your Azure subscription. AML acts solely as a processing platform.
You can export any artifact at any time using Azure Storage Explorer, Azure CLI, or the portal.
Compliance & Certifications
Azure Machine Learning complies with ISO/IEC 27001, SOC 1/2/3, GDPR, HIPAA BAA, and FedRAMP High. See the Azure compliance offerings for details.
You can provision AML workspaces in any Azure region, ensuring data never leaves the geographic boundary you select.
Frequently Asked Questions
All data at rest is encrypted using Azure Storage Service Encryption (AES‑256). Data in transit uses TLS 1.2 or higher.
Yes. Azure Key Vault integration allows you to supply customer‑managed keys for storage encryption.
Submit a support ticket through the Azure portal. Include the workspace ID and the resources you deleted. Azure will provide a compliance report.