AI Policy Log

Tool page

Microsoft Copilot

Microsoft’s official pages currently split Microsoft 365 Copilot, the updated consumer Copilot app (as of August 18, 2026), and an older consumer FAQ that Microsoft says no longer applies after that update. We quote each page separately rather than collapsing them into one rule.

Checked 2026-11-18 (America/Toronto). Page status: complete · Vendor: Microsoft

Not a verdict. These are quotations from official vendor pages. They are not legal advice and not a determination that a plan is safe, compliant, or suitable for a particular kind of data.

Policy by plan

Plan Trains on inputs by default? Opt-out / opt-in Retention ZDR BAA / DPA Data residency Exceptions
Microsoft 365 Copilot (work or school / Microsoft 365 apps) No, according to Microsoft’s Copilot privacy documentation: “Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs, including those used by Microsoft Copilot.” Optional customer feedback is also described as not used to train those foundation LLMs. The pages we read present this as a default commitment, not a user toggle. Admins can manage feedback through admin controls. Admins can use Microsoft Purview to set retention policies for data related to chat interactions with Copilot, and can use Content search or Purview to view and manage stored data. Not stated / not verified Microsoft 365 commercial customers have Microsoft’s Data Protection Addendum / product terms; we did not independently extract a Copilot-specific BAA sentence from the pages in this pass, so a Copilot-only BAA statement is not verified here. Not stated / not verified in the excerpts we captured from the privacy article (Microsoft 365 residency is a broader product topic). This article does not apply when you use the consumer Copilot app signed in with a personal Microsoft account. Connected-experience personal data in Microsoft 365 is also described as not used to train LLMs, including those used by Microsoft Copilot.
Microsoft Copilot app for individuals (updated app as of August 18, 2026), signed in with a personal Microsoft account No, on the current activity-history article: “Prompts, responses, and your file contents when using the Microsoft Copilot app aren't used to train foundation models.” Optional feedback may be used to improve Copilot but “We don't use this feedback to train the foundation models used by Copilot.” Presented as a product rule on this article, not as a training opt-out toggle. Privacy controls for personalization and deleting chats are documented separately. Not stated / not verified as a numbered default period on this activity-history article (it points to chat-history and privacy-dashboard articles for deletion and export). Not stated / not verified Not stated / not verified Not stated / not verified This article “does not apply when you sign in … with a work, school, or organizational (Microsoft Entra) account,” and does not apply to Copilot in Microsoft 365 apps (see the home/personal M365 article Microsoft links). If you have not updated to the new Copilot app, Microsoft says this article does not apply and points to the older Privacy FAQ.
Older Microsoft Copilot app (Privacy FAQ; Microsoft says this applies only if you have not updated to the August 18, 2026 app) This FAQ, which Microsoft labels as applying only to the older app, said Microsoft used conversation activity from some consumer Copilot users to train generative AI models unless they opted out, and listed categories of users excluded from training (including Entra ID / organizational accounts, Copilot in Microsoft 365 Personal or Family, signed-out users, users under 18, opted-out users, and users in listed countries). The older FAQ described an opt-out from use of conversation activity for model training. Do not treat this as the rule for the updated app. Not stated / not verified on the FAQ excerpts we captured. Not stated / not verified Not stated / not verified Not stated / not verified Microsoft 365 Copilot conversations “remain read-only for you and are not used to train Copilot’s generative AI models,” even in this older FAQ. Read the FAQ only if you are still on the older app, as Microsoft instructs.

Source quotes

“Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs, including those used by Microsoft Copilot.” Microsoft 365 Copilot (work or school / Microsoft 365 apps) · Checked 2026-11-18 · Microsoft 365 Copilot privacy (Microsoft Learn)
“Prompts, responses, and your file contents when using the Microsoft Copilot app aren't used to train foundation models. We may use customer feedback, which is optional, to improve Microsoft Copilot … We don't use this feedback to train the foundation models used by Copilot.” Microsoft Copilot app for individuals (updated app as of August 18, 2026), signed in with a personal Microsoft account · Checked 2026-11-18 · Microsoft Copilot for individuals: your activity history
“An updated version of the Microsoft Copilot app for web, desktop, and mobile devices is available as of August 18, 2026. The information in this article applies only to the older version of the Microsoft Copilot app. … You can opt out of use of your conversation activity for model training at any time.” Older Microsoft Copilot app (Privacy FAQ; Microsoft says this applies only if you have not updated to the August 18, 2026 app) · Checked 2026-11-18 · Privacy FAQ for Microsoft Copilot (older app only)

Sources checked

Changelog for this tool

No approved later changes. First compiled 2026-11-18.