AI data policies

Microsoft (Azure OpenAI/Foundry, Copilot)

What Microsoft (Azure OpenAI/Foundry, Copilot) keeps, for how long, whether it trains on it, and what rights you grant, for Azure OpenAI / Foundry Models sold by Azure, Microsoft 365 Copilot (work/school accounts), Microsoft Copilot (consumer website and apps, personal Microsoft account). Each answer quotes Microsoft (Azure OpenAI/Foundry, Copilot)'s own policy.

Type Cloud platform Policies captured 2026-11-08 3 documents read How this is made

Azure OpenAI / Foundry Models sold by Azure API

Default retention How long prompts and chats are kept

Stated Plain inference is stateless and stores nothing in the model. Optional features (Responses API, Assistants Threads, Stored completions, Files) keep data in your resource until you delete it.

What the policies say (3 quotes)

"**The models are stateless: no prompts or completions are stored in the model. Additionally, prompts and completions are not used to train, retrain, or improve the base models.**"

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"This data is either uploaded by the customer, using the Files API or vector store, or is automatically stored in connection with certain stateful entities such as the Responses API, the Threads feature of the Assistants API, and Stored completions."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Can be deleted by the customer at any time."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
After you delete How long until it is gone

Stated Customers can delete stored data and fine-tuned models at any time. The page gives no timeframe for when deleted data is fully gone, including from backups.

What the policies say (2 quotes)

"Can be deleted by the customer at any time."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"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."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Kept longer when Safety flags, legal holds, backups

Stated Abuse monitoring: when abuse is indicated, sampled prompts and completions are kept in an abuse-monitoring data store for human review. This page doesn't say how long that store keeps them.

What the policies say (3 quotes)

"When these indicators are detected, a sample of customer's prompts and completions may be selected for review."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"The abuse monitoring data store where prompts and completions are stored for human review is logically separated by customer resource (each request includes the resource ID of the customer's Foundry resource)."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"For automated review, customer's prompts and completions are not stored by the system or used to train the AI models or other systems."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Zero data retention Whether it exists and who can get it

Stated Managed customers can apply for modified abuse monitoring. If approved, nothing is stored for human abuse review, though automated review may still run. ContentLogging=false confirms it.

What the policies say (3 quotes)

"As depicted in the diagram above, managed customers may [apply to modify abuse monitoring]"

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"If the customer has been approved for modified abuse monitoring (learn more at [Abuse Monitoring](/en-us/azure/ai-foundry/openai/concepts/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."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"The value of "false" for the "ContentLogging" attribute appears only if data storage for abuse monitoring is turned off."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Trains on your data By default

Stated No. Prompts, completions, embeddings and training data aren't used to train foundation models or improve Microsoft or third-party products without your permission, and aren't shared with OpenAI.

What the policies say (2 quotes)

"are NOT used to train any generative AI foundation models without your permission or instruction."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Customer Data, Prompts, and Completions are NOT used to improve Microsoft or third-party products or services without your explicit permission or instruction."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Opting out How, and what it doesn't cover

Stated No opt-out is needed: training happens only with your permission or instruction. Fine-tuned models from your data are yours alone. How permission is given isn't described.

What the policies say (2 quotes)

"Training data uploaded for fine-tuning is not used to train any generative AI foundation models without your permission or instruction."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Your fine-tuned Models sold by Azure are available exclusively for your use."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Human review Whether people read conversations

Stated Possible by default: automated review, plus humans 'as necessary'. Authorized Microsoft staff may review flagged prompts and completions under controlled access. EEA deployments use EEA-based reviewers.

What the policies say (3 quotes)

"Review is conducted by automated means including by AI models such as LLMs by default, with additional reviews by human reviewers as necessary."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Human reviewers assessing potential abuse can access prompts and completions data only when that data has already been flagged by the abuse monitoring system, or when the prompts and completions are part of a potentially abusive pattern of use."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"For Models sold by Azure deployed in the European Economic Area, the authorized Microsoft employees are located in the European Economic Area."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Memory Remembered facts and profiles

Stated No end-user memory feature is described. Stateful features (Responses API, Assistants Threads, Stored completions) save message history in your resource, configured by you and deletable.

What the policy says

"When you use certain optional features of Models sold by Azure and Agents, such as the [Responses API](/en-us/azure/ai-foundry/openai/how-to/responses), the Threads feature of the [Assistants API](/en-us/azure/ai-foundry/openai/how-to/assistant), and Stored completions, the service creates a data store to persist message history and other content, in accordance with how you configure the feature."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Connected apps Email, calendar, drive

Stated With "On your data", data stays in the source you connect and the service doesn't copy it into its own store. What happens on disconnect isn't stated.

What the policy says

"The data remains stored in the data source and location you designate; Azure OpenAI does not create a duplicate data store."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Agent actions Actions taken, browsing, credentials

Stated Agents' stateful features keep message history in a data store, and prompts may pull in content from URLs. Logs of agent actions, browsing and stored credentials aren't covered.

What the policies say (2 quotes)

"When you use certain optional features of Models sold by Azure and Agents, such as the [Responses API](/en-us/azure/ai-foundry/openai/how-to/responses), the Threads feature of the [Assistants API](/en-us/azure/ai-foundry/openai/how-to/assistant), and Stored completions, the service creates a data store to persist message history and other content, in accordance with how you configure the feature."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Prompts may also be augmented with data retrieved from a source included in the prompt itself, such as a URL."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Your inputs Who owns them

Not stated The captured policies don't say who owns inputs. That is likely in the Product Terms or DPA, which aren't captured.

Outputs Who owns them

Not stated The Azure page doesn't say who owns outputs.

License you grant Scope, and whether it is perpetual

Not stated No license grant from the customer is described in the captured policies.

Sharing Model providers, subprocessors, ads

Stated Data isn't available to OpenAI, other model providers or other customers. Models run in Microsoft's Azure environment and don't call OpenAI's services. The Microsoft DPA governs processing.

What the policies say (3 quotes)

"are NOT available to OpenAI or other providers of Models sold by Azure."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Foundry is an Azure service; Microsoft hosts the Models sold by Azure in Microsoft's Azure environment and Models sold by Azure do NOT interact with any services operated by providers of Models sold by Azure, for example, OpenAI (e.g. ChatGPT, or the OpenAI API)."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"Please also see [Microsoft Products and Services Data Protection Addendum](https://aka.ms/DPA), which governs data processing by Models sold by Azure."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Where data lives Regions and residency

Stated Standard deployments process in your chosen geography. Global may process wherever the model runs; DataZone stays within the US or EU zone. Data at rest stays in your geography.

What the policies say (2 quotes)

"Prompts and responses are processed within the customer-specified [geography](https://azure.microsoft.com/explore/global-infrastructure/geographies/) (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)."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"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."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08
Enterprise controls Retention settings, DPA, SSO

Stated Covered by the Microsoft DPA. Stored data can use customer-managed keys, and managed customers can apply for modified abuse monitoring. SSO and audit logs aren't covered on this page.

What the policies say (2 quotes)

"Is always encrypted at rest with Microsoft's AES-256-encryption by default, with the option of using a customer managed key (certain preview features may not support customer-managed keys)."

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

"As depicted in the diagram above, managed customers may [apply to modify abuse monitoring]"

Data, privacy, and security for Foundry Models sold by Azure in Microsoft Foundry - Microsoft Foundrylearn.microsoft.comcaptured 2026-11-08

Microsoft 365 Copilot (work/school accounts) Business

Default retention How long prompts and chats are kept

Stated Prompts, responses and citations are stored as Copilot activity history under the same contractual commitments as your org's other Microsoft 365 content. No default retention period is stated.

What the policies say (2 quotes)

"The stored data includes the user's prompt and Copilot's response, including citations to any information used to ground Copilot's response."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"This data is processed and stored in alignment with contractual commitments with your organization's other content in Microsoft 365. The data is encrypted while it's stored and isn't used to train foundation LLMs, including those used by Microsoft Copilot."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
After you delete How long until it is gone

Stated Users can delete their Copilot activity history in the My Account portal. No timeframe for final deletion is given.

What the policy says

"Your users can delete their Copilot activity history, which includes their prompts and the responses Copilot returns, by going to the [My Account portal](https://myaccount.microsoft.com/)."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Kept longer when Safety flags, legal holds, backups

Stated Admins can set Purview retention policies on Copilot chat data and search it. Copilot has opted out of Azure OpenAI abuse monitoring. Legal-hold and backup durations aren't stated.

What the policies say (2 quotes)

"To view and manage this stored data, admins can use Content search or Microsoft Purview. Admins can also use Microsoft Purview to set retention policies for the data related to chat interactions with Copilot."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"While abuse monitoring, which includes human review of content, is available in Azure OpenAI, Microsoft Copilot services have opted out of it."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Zero data retention Whether it exists and who can get it

Inferred No zero-retention option is described. Inferred from the page saying interactions are stored as activity history by default.

What the policy says

"When a user interacts with Microsoft Copilot (using apps such as Word, PowerPoint, Excel, OneNote, Loop, or Whiteboard), we store data about these interactions."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Trains on your data By default

Stated No. Prompts, responses and Microsoft Graph data aren't used to train foundation LLMs.

What the policy says

"Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs, including those used by Microsoft Copilot."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Opting out How, and what it doesn't cover

Stated No training opt-out is needed. Optional user feedback may be used to improve Copilot, though not to train foundation LLMs, and admins can manage feedback.

What the policy says

"We may use customer feedback, which is optional, to improve Microsoft Copilot, just like we use customer feedback to improve other Microsoft 365 services and Microsoft 365 productivity apps. We don't use this feedback to train the foundation LLMs used by Microsoft Copilot. Customers can manage feedback through admin controls."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Human review Whether people read conversations

Stated Microsoft Copilot services have opted out of Azure OpenAI abuse monitoring, which includes human review. Automated harmful-content filters still apply.

What the policy says

"While abuse monitoring, which includes human review of content, is available in Azure OpenAI, Microsoft Copilot services have opted out of it."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Memory Remembered facts and profiles

Stated Only Copilot activity history is described: users can delete it, and admins can view it via Content search or Purview. A separate memory feature isn't mentioned.

What the policies say (2 quotes)

"We refer to the user's prompt and Copilot's response to that prompt as the "content of interactions" and the record of those interactions is the user's Copilot activity history."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"To view and manage this stored data, admins can use Content search or Microsoft Purview."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Connected apps Email, calendar, drive

Stated Copilot uses Microsoft Graph content (email, calendar, chats, documents) the user can already view, plus Graph connectors for third-party data. What happens to indexed data on disconnect isn't stated.

What the policies say (3 quotes)

"It can generate responses anchored in your organizational data, such as user documents, emails, calendar, chats, meetings, and contacts."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Microsoft Copilot only surfaces organizational data to which individual users have at least view permissions."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Data from Graph connectors can be returned in Microsoft Copilot responses if the user has permission to access that information."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Agent actions Actions taken, browsing, credentials

Stated Copilot sends queries built from prompts, activity history and accessible data to admin-approved agents, which follow their own terms. Web queries go to Bing. Action logs aren't described.

What the policies say (3 quotes)

"If an agent is needed, Microsoft Copilot generates a search query to send to the agent on the user's behalf. The query is based on the user's prompt, Copilot activity history, and data the user has access to in Microsoft 365."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"When you're using agents to help Microsoft Copilot to provide more relevant information, check the privacy statement and terms of use of the agent to determine how it will handle your organization's data."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Based on these terms, Copilot generates a search query that it sends to the Bing Search service."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Your inputs Who owns them

Not stated The page doesn't say who owns inputs. It says only that they are handled like other Microsoft 365 content under the org's contract.

Outputs Who owns them

Stated Microsoft doesn't claim ownership of output, but doesn't decide whether it is copyrightable. It will defend commercial customers sued for copyright if they used the built-in guardrails.

What the policies say (2 quotes)

"Microsoft doesn't claim ownership of the output of the service. That said, we don't make a determination on whether a customer's output is copyright protected or enforceable against other users."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"If a third party sues a commercial customer for copyright infringement for using Microsoft's Copilots or the output they generate, we'll defend the customer and pay the amount of any adverse judgments or settlements that result from the lawsuit, as long as the customer used the guardrails and content filters we have built into our products."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
License you grant Scope, and whether it is perpetual

Not stated No license grant from the customer is described in the captured policies.

Sharing Model providers, subprocessors, ads

Stated Admins can choose to use Anthropic and OpenAI models, which act as subprocessors (extra terms may apply). Other Microsoft-hosted models keep data inside Microsoft. Web search queries go to Bing.

What the policies say (2 quotes)

"Microsoft offers third-party AI models in Microsoft Copilot, such as Anthropic and OpenAI models. As an admin, you can decide whether to use these models to power your experiences, and additional terms may apply."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"These models are governed by the same contractual and data protection commitments already in place, including that no data leaves Microsoft."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Where data lives Regions and residency

Stated EU traffic stays in the EU Data Boundary, except Anthropic models. Non-EU queries may go to the US, EU or other regions. Advanced Data Residency and Multi-Geo cover Copilot.

What the policies say (3 quotes)

"EU traffic stays within the EU Data Boundary while worldwide traffic can be sent to the EU and other countries or regions for LLM processing."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Models provided by Anthropic as a subprocessor are currently excluded from the EU Data Boundary."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Microsoft [Advanced Data Residency (ADR)](/en-us/microsoft-365/enterprise/m365-dr-adr-overview) and [Multi-Geo Capabilities](/en-us/microsoft-365/enterprise/microsoft-365-multi-geo) offerings include data residency commitments for Microsoft Copilot customers as of March 1, 2024."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08
Enterprise controls Retention settings, DPA, SSO

Stated Product Terms and DPA apply. Admins get Purview retention policies, Content search, Teams Export APIs, agent allow-lists, feedback controls and connected-experience privacy controls.

What the policies say (3 quotes)

"Microsoft Copilot is upholding data residency commitments as outlined in the Microsoft Product Terms and Data Protection Addendum."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"For Microsoft Teams chats with Copilot, admins can also use [Microsoft Teams Export APIs](/en-us/microsoftteams/export-teams-content) to view the stored data."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

"Admins have full control to select which agents are allowed in their organization."

Data, Privacy, and Security for Microsoft Copilotlearn.microsoft.comcaptured 2026-11-08

Microsoft Copilot (consumer website and apps, personal Microsoft account) Consumer app

Default retention How long prompts and chats are kept

Stated No Copilot-specific retention period. The privacy statement says data is kept as long as necessary for the service and other legitimate purposes, and periods vary widely.

What the policy says

"Because these needs can vary for different data types and the context of our interactions with you or your use of our products, actual retention periods can vary significantly. See relevant product documentation for more information."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
After you delete How long until it is gone

Stated Signed-in users can view, export and delete Copilot activity history on the privacy dashboard. No timeline is given. The only closure timeline (60 days) is stated for child accounts.

What the policies say (2 quotes)

"If you are signed into Microsoft Copilot with your Microsoft account, you can view, access, export and delete Copilot activity history from your [privacy dashboard](https://account.microsoft.com/privacy/)."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"After requesting account closure, the account is marked for deletion, but we will wait 60 days before permanently deleting the account in case you change your mind or need to access information."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Kept longer when Safety flags, legal holds, backups

Stated Data may be kept for legal duties, preservation orders, litigation, security and safety. Content can be preserved and disclosed under good-faith legal or safety needs. No durations are given.

What the policies say (2 quotes)

"Legal, contractual, and other similar obligations we are subject to, including data retention laws, government orders to preserve data relevant to an investigation, or data retained for the purposes of litigation."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"Finally, we retain, access, transfer, disclose, and preserve personal data, including your content (like the content of your emails in Outlook.com or files in OneDrive), when we have a good faith belief that it is necessary to do any of the following:"

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Zero data retention Whether it exists and who can get it

Not stated No zero-retention option is described for consumer Copilot.

Trains on your data By default

Stated Microsoft may use data to train AI models, and Copilot uses prompts to improve services. Whether Copilot chats are trained on by default isn't stated. Copilot Health is excluded.

What the policies say (3 quotes)

"As part of our efforts to improve and develop our products, we may use your data to develop, train, and fine-tune our AI models, including large language models (LLMs)."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"Microsoft Copilot also uses prompts and related data to provide and improve services, including relevant advertising."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"It is not used to develop or train generative AI models, and it is not used for advertising or marketing purposes."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Opting out How, and what it doesn't cover

Not stated The privacy statement describes no Copilot training opt-out. It links to a Copilot privacy FAQ that isn't in the captured policies.

Human review Whether people read conversations

Inferred Likely, but inferred: Microsoft says it manually reviews some AI results and uses conversation data to prevent abuse. It doesn't say whether humans read Copilot chats specifically.

What the policies say (2 quotes)

"Our processing of personal data involves both automated and manual (human) methods, which often work together. For example, to build, train, and improve the accuracy of our automated systems – such as AI - we manually review some of the results against the underlying data."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"We also use your conversation data to monitor performance, troubleshoot and fix bugs and issues, prevent abuse, and to provide and improve Microsoft Copilot."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Memory Remembered facts and profiles

Stated In some markets, Copilot uses conversation history and saved memories to personalize, and may draw on Bing, Edge and MSN activity. You can manage memories and personalization in Copilot settings.

What the policy says

"For users in some markets, Microsoft Copilot can use your conversation history (including any memories you save) to personalize your experience. We may also offer you the ability to personalize your Microsoft Copilot experience using your interactions with our other products and services, like Bing, Edge, and MSN. You can manage your memories and control personalization features using the settings in Microsoft Copilot at any time."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Connected apps Email, calendar, drive

Stated In Microsoft 365 Personal/Family/Premium apps, Copilot uses file content only when you ask it to act on that content. Email or calendar connectors and disconnect behavior aren't described.

What the policy says

"When used with available Microsoft 365 apps, Microsoft Copilot will use the content in your files only when you ask it to perform a specific action on that specific content"

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Agent actions Actions taken, browsing, credentials

Stated Copilot can act in other apps and sites. In Edge, it may see browsing activity if you allow it, and captured policies pages it visits during tasks. Retention isn't stated.

What the policies say (3 quotes)

"as well as let Microsoft Copilot interact with other apps, services, and websites to take actions on your behalf."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"If you choose to use features that perform tasks on your behalf, Copilot will capture visual snapshots of the pages it visits to interact with and complete the action."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"When using Copilot features in Microsoft Edge, you can choose to allow Microsoft Copilot in Edge to access your browsing activity, including screenshots of webpages you visit, open tabs, and browsing history to provide better answers and enable advanced Copilot features."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Your inputs Who owns them

Not stated The privacy statement doesn't say who owns inputs. The Microsoft Services Agreement, which isn't captured, would likely cover it.

Outputs Who owns them

Not stated Output ownership for consumer Copilot isn't stated in the captured policies.

License you grant Scope, and whether it is perpetual

Not stated No content license is described in the captured policies.

Sharing Model providers, subprocessors, ads

Stated Ads may relate to your Copilot conversations, including shared files. Data goes to ad partners for personalized ads (opt-out available) and to vendors. Third-party platforms' policies may also apply.

What the policies say (3 quotes)

"Ads may be shown that relate to the current interaction you are having with us, such as your Copilot conversations (including files you share); your current location; transactions; product usage; search queries; or the content you’re viewing."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"To opt out of sharing data with third parties for personalized ads, please visit our [third-party ad settings](https://go.microsoft.com/fwlink/?linkid=2214802) page."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

"If you use Microsoft Copilot in third-party products and services, such as messaging platforms, Microsoft processes your data as described here, but the third party’s privacy policies and data processing activities may also apply."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Where data lives Regions and residency

Stated Data may be stored in your region, the US, or wherever Microsoft or its providers operate, typically your region or the US with a backup elsewhere. No residency choice.

What the policy says

"Typically, the primary storage location is in your region or in the United States, often with a backup to a data center in another region."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08
Enterprise controls Retention settings, DPA, SSO

Stated This is a consumer product, so no admin controls apply. The statement points enterprise customers to its Enterprise and Developer Products section.

What the policy says

"For enterprise customers, Microsoft Copilot offers enhanced data protection and customization. Data is collected and used as described in the Enterprise and Developer Products section."

Microsoft Privacy Statement \u2013 Microsoft privacymicrosoft.comcaptured 2026-11-08

Notes

Open questions

Questions the captured policies leave unanswered, and where the answer probably lives.

Documents read