Microsoft will switch on Claude in Copilot for EU tenants on 9 December

Microsoft will switch on Claude in Copilot for EU tenants on 9 December

10 November 2026: Microsoft will enable Claude models in Copilot for all users in EU and EFTA tenants that have not made a choice, starting on 9 December, reversing today's default of off. An administrator who wants them kept off needs to record that decision before the date.

Not yet independently verified. We read this notice through a public archive of Microsoft 365 Message Center posts, because the original sits inside the tenant admin console. Microsoft does not state in the post which tenants count as eligible beyond naming the EU and EFTA. We will update this when it can be confirmed, and remove this note.

Moving into the EU Data Boundary a few days earlier

The change is tied to data location. Claude models are not covered by Microsoft's EU Data Boundary processing commitment today, which is why they have been disabled by default in the region. From early December, processing for eligible tenants moves inside that boundary, and the default flips once it has.

Anthropic stays a Microsoft subprocessor. Microsoft says Claude models in Copilot fall under its enterprise data protection, and that the Product Terms and Data Protection Addendum apply "except as otherwise disclosed in Microsoft Learn documentation". A data protection officer will want that exception read before the switch. Anthropic models with data retention are governed by a separate setting and are not part of this change.

How each existing selection is treated

The outcome depends on the setting a tenant holds under "AI providers operating as Microsoft subprocessors":

• No selection: Claude models are enabled for all users on 9 December.
• Claude allowed only in Copilot in Microsoft 365 apps: that option is retired, and the tenant moves to All users.
• All users, or specific users and groups: assignments stay, and processing moves into the boundary automatically.
• No users: the selection is respected.

The second row is the easy one to miss. A tenant that deliberately chose the narrower option is widened by this change, across Copilot experiences and Copilot Studio, unless someone changes the setting first.

The setting to change before the date

The control is in the Microsoft 365 admin center under Copilot, Settings, View all, AI providers operating as Microsoft subprocessors, Anthropic. Select No users to opt out or Specific users and groups to limit access, then save. It needs the AI Administrator or Global Administrator role. Admins can still disable the models at any point after 9 December, so missing the date is recoverable, though users will have had access in the meantime.

Microsoft tags the post as a major change with an act-by date of 9 December and asks admins to tell their helpdesk and compliance teams. Organisations with a records of processing entry or a subprocessor register naming the AI providers in use should update it, since a new subprocessor in active use and a new processing location are both things those documents exist to capture.

Sources

• Microsoft 365 Message Center MC1546438 (public archive copy), 10 November 2026