Microsoft will automatically enable access to Claude in Copilot for British companies

Microsoft announced changes to the access rules for Anthropic's Claude AI models in corporate Copilot products. Starting December 9, 2026, the models will be automatically enabled in Microsoft 365 organizations in the UK unless their administrators have previously opted out. This was stated in a Microsoft 365 Message Center notification published on November 10.

The change will affect companies using Microsoft 365 Copilot and related corporate tools. However, Claude will not be a mandatory model for every request: Microsoft will expand its use, and the choice of a specific model will depend on the product and task at hand. Administrators will retain the ability to completely prohibit access to Anthropic or grant it only to certain employees.

What Claude settings will change on December 9

Currently, for corporate organizations in the UK, the use of Claude models in Microsoft Copilot is disabled by default. Administrators can independently enable them through the Microsoft 365 admin center by selecting Anthropic from the list of AI providers acting as Microsoft's data subprocessors.

From December 9, companies that have not made such a decision will automatically gain access to Claude for all users. If an organization has already prohibited Anthropic models or allowed them for specific users and groups, existing settings will remain. Claude can be disabled or the list of users changed even after the specified date.

Microsoft will also cancel a separate mode where Claude was allowed to be used only within Microsoft 365 applications, including Word, Excel, and PowerPoint. Organizations with such a setting will be switched to a broader All users mode, covering supported Copilot capabilities. To maintain restricted access, administrators will need to select specific user groups or completely disable Anthropic.

The changes will expand the availability of Claude in Microsoft's corporate tools, including Copilot, Researcher, and Copilot Studio. However, some capabilities require additional settings, and certain Anthropic models, for which data storage by the provider is provided, will remain under independent administrative control.

Where corporate data will be processed

One of the main concerns for UK companies is the location of the infrastructure processing requests to Claude. According to Microsoft documentation, when using Anthropic models through the corresponding generative AI services, client data is processed in the US. Anthropic itself acts as a data subprocessor for Microsoft and operates within the contractual and technical requirements of the corporate platform.

For the UK, this is important because the country is not part of the EU Data Boundary regional zone. This program includes Microsoft's commitments to store and process certain client data within the European Union and the European Free Trade Association. Therefore, UK organizations need to separately assess the conditions of international data transfer when using Claude.

Microsoft emphasizes that standard Claude models in corporate Copilot products fall under its Enterprise Data Protection mechanisms and corresponding contractual obligations. According to the company, client data is not used to train foundational generative AI models. However, specific conditions may differ for individual models requiring additional administrator consent for data storage.

Changes will also affect EU companies

At the same time, Microsoft is preparing a similar change for organizations in the European Union and European Free Trade Association countries. According to a separate notification from November 10, on December 9, access to Claude will also be automatically enabled for relevant clients if their administrators have not previously chosen Anthropic settings.

For these regions, an additional change is planned: in early December, Claude data processing should transition to infrastructure covered by the EU Data Boundary. Microsoft announced this possibility on November 5, explaining the need to provide European companies with a choice of AI models without abandoning regional data processing commitments.

Thus, the changes for the UK and the EU coincide in date but differ in data placement and processing conditions. Organizations operating in multiple countries will need to consider these differences when configuring corporate access to Anthropic models.

What administrators need to do

Microsoft recommends that UK companies check Claude settings by December 9. To do this, open the Microsoft 365 admin center, go to the Copilot section, then in general settings find AI providers acting as Microsoft subprocessors, and select Anthropic.

If an organization does not want to allow Claude for all employees, the administrator can select No users for complete disabling or Specific users and groups to grant access to certain users. Changing settings requires the role of AI Administrator or Global Administrator.

If the current configuration already meets the company's requirements, no additional actions will be needed. However, organizations that previously restricted Claude only to Microsoft 365 applications should check the settings separately, as this mode will be canceled on December 9.