Microsoft 365 Copilot: Admins will be able to enable third‑party model providers for specific users and groups

Admins will soon be able to assign third-party AI model providers like Anthropic and xAI to specific users or groups in Microsoft 365 Copilot via a new control in the Microsoft Admin Center. This rollout begins in late April 2026 and supports Entra ID group-based access management.

[Introduction]

Today, IT admins can enable or disable third‑party model providers, such as Anthropic and xAI (US only), at the tenant level in the Microsoft Admin Center. With this update, admins will be able to assign access to a third‑party model provider to specific users or groups within the tenant.

This update is related to MC1193920, which announced that Anthropic models would be enabled by default for many customers in certain Copilot experiences and offered under the Microsoft Product Terms and Data Protection Addendum (DPA) as a Microsoft subprocessor.

This message corresponds to Microsoft 365 Roadmap ID 557371.

[When this will happen]

General Availability (Worldwide): We will begin rolling out in late April 2026 and expect to complete in late April 2026.

[How this affects your organization]

Who is affected

• Organizations that use or plan to use third‑party model providers in Microsoft 365 Copilot and in Copilot Studio
• Admins who manage access in the Microsoft Admin Center, Power Platform Admin Center (PPAC), or Copilot Studio

What will happen

• Admins will see a new control in the Microsoft Admin Center that allows assigning a third‑party model provider to specific users or Entra ID groups.
• The setting will apply at the provider level, not at the individual model level.
• Assignments will be enforced consistently across the Microsoft Admin Center, PPAC, and Copilot Studio.
• The control will apply to subprocessors and independent processors for all current and future third‑party model providers.
• Admins can add up to any combination of 999 groups and users; nested groups are supported. 

[What you can do to prepare]

• Review your existing settings for Anthropic and xAI (US only) in the Microsoft Admin Center.
• Update user or group assignments to align with your internal governance policies.
• Update any existing agents and experiences that may be impacted when model access is limited by Entra ID security groups.
• Communicate any planned changes to your helpdesk or impacted user groups.
• Update internal documentation where you track AI access or model provider settings.
• Review Microsoft documentation on managing third‑party model providers when it becomes available.

Learn more:

• Anthropic as a subprocessor for Microsoft Online Services | Microsoft 365 Copilot | Microsoft 365 | Microsoft Learn
• Choose an external model as the primary AI model | Microsoft Copilot Studio | Microsoft Learn
• Enterprise data protection in Microsoft 365 Copilot and Microsoft 365 Copilot Chat | Microsoft 365 Copilot | Microsoft 365 | Microsoft Learn
• Microsoft Products and Services Data Protection Addendum (DPA) | Licensing Resources and Documents | Licensing | Microsoft
• Product Terms | Licensing | Microsoft

[Compliance considerations]

Question	Answer
Does the change introduce or significantly modify AI, ML, or agent capabilities that interact with or provide access to customer data?	Yes. This update introduces new admin controls that define which users or groups can access third‑party AI model providers.
Does the change add any integration to third‑party software products, and if so what?	Yes. This update applies to model providers such as Anthropic and xAI (US only).
Does the change include an admin control and, can it be controlled through Entra ID group membership?	Yes. Access can be assigned at the user or group level using Entra ID.