Data retention - verified
Does Azure OpenAI store your API prompts?
Default retention: flagged content only; duration not stated. Trains on your API data: No. Zero data retention: enterprise only (modified abuse monitoring, for Microsoft-managed customers, by application). Regional processing: US, EU and APAC.
Scope: Foundry Models sold by Azure, including Azure OpenAI models. Claude in Foundry is covered under Anthropic's terms, not this page.
Read from learn.microsoft.com. Compare every AI API's data retention
- Trains on data
- No
- ZDR
- Enterprise only
- Regions
- US, EU and APAC
- Verified
- 2026-10-18
How long are prompts kept by default?
- Default retention
Flagged content only; duration not stated
Prompts and completions flagged by abuse monitoring may be stored for human review in an abuse monitoring data store in the resource's geography. The current Microsoft Learn pages fetched do not state a retention duration in days.
“A separate data store is located in each geography in which the model sold by Azure is available, and a customer's prompts and generated content are stored in the Azure geography where the customer's Foundry resource is deployed, within the Models sold by Azure service boundary.”
learn.microsoft.com “For automated review, customer's prompts and completions are not stored by the system or used to train the AI models or other systems.”
learn.microsoft.com “The models are stateless: no prompts or completions are stored in the model.”
learn.microsoft.com - Duration
- Not stated
- Purpose
abuse monitoring (Code of Conduct / product terms)
Does it train on your API data?
No
“are NOT used to train any generative AI foundation models without your permission or instruction.”
“are NOT available to OpenAI or other providers of Models sold by Azure.”
Zero data retention (ZDR)
- Availability
Enterprise only
Modified abuse monitoring, for Microsoft-managed customers, by application
Recorded as: No product called ZDR is documented; the equivalent is modified abuse monitoring, which turns off abuse-monitoring data storage and human review (automated review may still run).
- Who can get it
Limited Access: only customers and partners managed by a Microsoft account team or under an eligible program, by application form. Some advanced models may have stricter criteria.
- Not covered
Stateful features (Responses API, Assistants threads, stored completions, Files, vector stores, batch, fine-tuning data) still store data in the customer's resource until deleted. Preview features may follow different practices.
- From the docs
“At this time, modified Guardrails (previously content filters) and/or modified abuse monitoring for Models sold by Azure are available only to customers and partners managed by a Microsoft account team or under an eligible program, and are subject to additional requirements.”
learn.microsoft.com “If the customer has been approved for modified abuse monitoring (learn more at Abuse Monitoring)), the data storage and human review process described above is not performed. However, automated review may still be conducted, leveraging algorithms including AI models that review prompts and completions at the time provided or generated, as applicable.”
learn.microsoft.com “Some advanced Models sold by Azure may have more stringent criteria for turning off abuse monitoring.”
learn.microsoft.com
Features kept longer
Stateful features hold your content until a job finishes, a window expires or you delete it, often regardless of the default above.
| Feature | Retention | The docs say |
|---|---|---|
| Responses API | 30 days by default | “By default, response data is retained for 30 days.”learn.microsoft.com |
| Batch job records | up to 365 days after terminal state | “Batch job records are retained for up to 365 days after the job reaches a terminal state, such as completed, failed, expired, or cancelled.”learn.microsoft.com |
| Batch output and error files | No default expiration; optional 14 to 30 days | “Generated output and error files don't have a default expiration. They remain available until you delete them or set an explicit expiration.”learn.microsoft.com |
| Files API, vector stores, Assistants threads, stored completions (preview) | Not stated; stored in customer's resource and deletable at any time | “Can be deleted by the customer at any time.”learn.microsoft.com |
| Fine-tuned models and training data | Until customer deletes | “Fine-tuned models are exclusively available to the customer whose data was used to create the fine-tuned model, are encrypted at rest (when not deployed for inferencing), and can be deleted by the customer at any time.”learn.microsoft.com |
Where is your data stored and processed?
- Regional processing
US, EU and APAC
Data Zones, or one Azure geography with Standard deployments
- Stated location
The resource's Azure geography, unless Global or Data Zone
- Regions
- Customer-specified Azure geography (Standard / Regional Provisioned)
- Data Zone US
- Data Zone EU (Azure EU Data Boundary)
- Data Zone APAC
- Global (any Azure region)
- How to choose
Deployment type: Standard or Regional Provisioned (Azure geography), Data Zone Standard/Provisioned/Batch (US, EU or APAC data zone), Global Standard/Provisioned/Batch (any region). Resource location sets the storage geography.
- What it covers
Storage at rest stays in the resource's geography for all deployment types (including abuse monitoring store for Global/DataZone); deployment type changes only processing location.
- From the docs
“Data residency for all deployment types: Data stored at rest remains in the designated Azure geography.”
learn.microsoft.com “Data Zone types: The service processes data only within the Microsoft-specified data zone (US, EU, or Asia Pacific (APAC)).”
learn.microsoft.com “The EU Data Zone follows the Azure EU Data Boundary, which can include European Free Trade Association (EFTA) countries and regions such as Norway and Switzerland in addition to EU member states. The APAC Data Zone covers multiple Asia Pacific regions. Microsoft can add regions to either data zone without prior notice to improve capacity and availability.”
learn.microsoft.com “For both Global and DataZone deployment types, any data stored at rest, such as uploaded data, and including the abuse monitoring data store created for Global and DataZone deployments, is stored in the customer-designated geography.”
learn.microsoft.com “Prompts and responses are processed within the customer-specified geography (unless you are using a Global or DataZone deployment type), but may be processed between regions within the geography for operational purposes (including performance and capacity management).”
learn.microsoft.com “Batch processing is a Global deployment type; data stored at rest remains in the designated Azure geography until processing capacity becomes available; processing may occur in any geography where the relevant model sold by Azure is deployed”
learn.microsoft.com
Regional endpoints
Regional hostnames: Not stated
Certifications
Certifications on the pages checked: Not stated
Research notes
- For EEA deployments, human reviewers are located in the EEA.
- ContentLogging=false in the resource JSON confirms abuse-monitoring storage is off.
- Stored data encrypted at rest with AES-256 by default, customer-managed keys optional.
- Abuse-monitoring retention duration (days) is not stated on the current data-privacy or abuse-monitoring pages.
FAQ
Does Azure OpenAI store your API prompts?
Flagged content only; duration not stated. Prompts and completions flagged by abuse monitoring may be stored for human review in an abuse monitoring data store in the resource's geography. The current Microsoft Learn pages fetched do not state a retention duration in days. The provider says: "A separate data store is located in each geography in which the model sold by Azure is available, and a customer's prompts and generated content are stored in the Azure geography where the customer's Foundry resource is deployed, within the Models sold by Azure service boundary."
Does Azure OpenAI train on your API data?
No. The provider says: "are NOT used to train any generative AI foundation models without your permission or instruction."
Does Azure OpenAI offer zero data retention?
Enterprise only (modified abuse monitoring, for Microsoft-managed customers, by application). Limited Access: only customers and partners managed by a Microsoft account team or under an eligible program, by application form. Some advanced models may have stricter criteria.
Where does Azure OpenAI process your data?
Regional processing options: US, EU and APAC (Data Zones, or one Azure geography with Standard deployments). Stated processing location: The resource's Azure geography, unless Global or Data Zone.
Sources
Last verified from Azure OpenAI's own pages below. Quotes are verbatim.
- https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy
- https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/abuse-monitoring
- https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/limited-access
- https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/responses
- https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/deployment-types
- https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/batch
This page summarizes what each provider's own pages say. It is not legal advice; your contract, data processing addendum and the provider's current terms govern what happens to your data.
Newsletter
Know when a provider changes its data terms
One email per week. New model releases, pricing and policy changes, and what they mean for what you build.