Get the weekly research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean. No account.
The statement authorizes Microsoft to use personal data collected through its products to develop, train, and fine-tune AI models including large language models. In some markets, users can opt out of Copilot conversation data being used for AI model training.
This analysis describes what Microsoft Azure's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
This provision establishes a broad authorization for AI training as a secondary use of personal data collected across Microsoft's consumer product portfolio, which may require evaluation under GDPR lawful basis requirements and U.S. state privacy laws governing secondary data processing purposes.
Interpretive note: The scope of the opt-out for AI training is stated to apply only in some markets and specifically to Copilot conversation data; the provision does not specify a universal opt-out for all AI training uses across all product data categories.
Microsoft now discloses that it may contact you by phone for marketing using automated dialers and AI-generated voices if you have consented to marketing communications, which represents a new disclosure of contact method and technology type. The company has also reorganized its data retention policy to state it retains data for broader business purposes including improving products and protecting systems, while removing previous specific examples and retention criteria, making it less clear exactly how long specific types of your data will be kept. You should review your consent settings for marketing communications and verify what contact methods you have authorized, particularly if you have concerns about automated or AI-generated calls.
View change record →Microsoft's privacy policy now provides a less detailed explanation of how long your data is retained. Previously, the policy included specific examples, such as how long deleted emails remain in your system before final deletion, and listed criteria for deciding retention periods. Now those details are consolidated into a more general statement pointing readers to separate product documentation. This means you'll need to consult multiple documents to understand retention timelines for specific services, which reduces transparency at the point of reading the main privacy policy.
View change record →Microsoft's updated retention policy provides greater specificity about how long your data persists and under what conditions it is deleted. The policy now explicitly states that deleted items from OneDrive and Outlook.com may remain in Microsoft's systems for up to 30 days before permanent removal, even after you empty the Deleted Items folder. Additionally, the updated terms clarify that retention periods depend on whether you have an expectation that Microsoft will keep the data until you actively remove it, and whether automated controls exist to let you access and delete data yourself. You can review Microsoft's privacy dashboard to exercise available deletion controls and understand which services retain your data under these criteria.
View change record →The agreement authorizes Microsoft to use personal data, including data collected during ordinary product use, to train and fine-tune AI models including large language models. The opt-out for Copilot conversation data used for AI training is stated to be available only in some markets, and the statement does not specify a universal opt-out mechanism for all AI training uses.
Cross-platform context
See how other platforms handle AI Model Training Authorization and similar clauses.
Compare across platforms →Monitoring
Microsoft Azure has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 25 platforms.
"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). Learn more here. Product development. We use data to develop new products. For example, we use data, often de-identified, to better understand our customers' computing and productivity needs, and to train and fine-tune AI models, which can shape the development of new products.Excerpt from Microsoft Azure's Microsoft Privacy
1. REGULATORY LANDSCAPE: This provision implicates GDPR Articles 5, 6, and 13 regarding lawful basis and purpose limitation for secondary processing, enforced by EU Data Protection Authorities. It also engages CCPA and applicable U.S. state privacy laws regarding secondary use of personal data and the distinction between service provision and product improvement purposes. The FTC Act's prohibition on unfair or deceptive practices is relevant to the adequacy of disclosure and consent mechanisms for AI training purposes. 2. GOVERNANCE EXPOSURE: High. The authorization to use consumer product data for AI model training as a secondary purpose creates compliance exposure under GDPR's purpose limitation principle, which requires that secondary processing be compatible with the original collection purpose or supported by a separate lawful basis such as consent. The geographic limitation of the Copilot training opt-out to some markets may create unequal protection across user populations. 3. JURISDICTION FLAGS: EU and EEA users face heightened exposure, as GDPR's purpose limitation and lawful basis requirements create constraints on secondary AI training use. UK GDPR raises parallel considerations enforced by the ICO. California residents may have rights under CCPA regarding secondary processing. The opt-out availability being market-dependent creates differential treatment across geographies. 4. CONTRACT AND VENDOR IMPLICATIONS: Enterprise customers whose employee or end-user data may be processed under consumer terms should review applicable enterprise agreements to confirm that AI training uses are governed by the enterprise agreement terms rather than this consumer privacy statement, given the statement's provision that enterprise agreement terms control in the event of conflict. Procurement teams should assess whether vendor agreements with Microsoft adequately restrict AI training use of organizational data. 5. COMPLIANCE CONSIDERATIONS: Legal teams should evaluate whether existing consent mechanisms and privacy notices adequately disclose AI training as a secondary processing purpose. Data mapping exercises should identify which data categories collected through consumer products may flow into AI training pipelines. Organizations subject to GDPR should assess whether a legitimate interests assessment or consent mechanism is required for this use, and whether current disclosures satisfy transparency obligations.
Full institutional analysis
Regulatory citations, enforcement risk, and due diligence action items.
Monitor: same-day alerts on the platforms you choose. Analyst: full institutional analysis.
Compliance Governance Intelligence
Need to monitor specific governance provisions?
Compliance includes provision-level monitoring, governance timelines, regulatory mapping, and audit-ready analysis.
Built from archived source documents, structured governance mappings, and historical version tracking.
This provision establishes a broad authorization for AI training as a secondary use of personal data collected across Microsoft's consumer product portfolio, which may require evaluation under GDPR lawful basis requirements and U.S. state privacy laws governing secondary data processing purposes.
The agreement authorizes Microsoft to use personal data, including data collected during ordinary product use, to train and fine-tune AI models including large language models. The opt-out for Copilot conversation data used for AI training is stated to be available only in some markets, and the statement does not specify a universal opt-out mechanism for all AI training uses.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Microsoft Azure.