Forum Discussion
Difference between Copilot Agent settings and Microsoft Agent 365 Agent settings
Hello,
I am trying to better understand the distinction between the agent governance settings available in the Microsoft 365 admin center.
Currently, I can find seemingly similar agent access settings in two different locations:
- Copilot → Settings → Agents
- Agents → Settings → User Access (Microsoft Agent 365)
For example, both locations appear to provide controls such as :
- Who can access agents
- Whether access is granted to all users or specific groups
- Agent usage permissions
From an administrator perspective, it is not clear whether these are two different configuration scopes, two different management experiences, or simply two entry points to the same underlying policy.
So :
- What is the intended difference between the Copilot settings experience and the Agent 365 settings experience ?
- Which interface should be considered the primary administration location going forward ?
- Are the settings synchronized and backed by the same configuration objects ?
- Is one of these experiences planned to replace the other in the future ?
- What governance scenarios should be managed in Copilot settings versus Agent 365 settings ?
Thank you for your help.
2 Replies
- sev7enBrass Contributor
Coralie, to make the verification suggested by Jamony concrete, I would use a small test matrix in a non-production scope: one user included in the access group, one excluded, and one agent assigned only to a restricted audience. Record separately whether each user can discover, install and invoke the agent, then test access to a permitted and a denied data source. Keep the agent, group membership and app policies fixed while changing one setting at a time through your normal change process. This helps identify which control is affecting the result without assuming the two screens share the same backend object.
In the governance model I published with PALO Framework (https://paloframework.org/PALO_AIGovernance.html), I connect each control to its scope, accountable owner, expected behaviour and observed evidence. For this case I would keep four rows: tenant access, individual agent assignment, data/tool permissions and spending eligibility. A successful access test in one row cannot validate the others.
Microsoft documents the centralized access control here: https://learn.microsoft.com/en-us/microsoft-365/admin/manage/agent-settings . If the two interfaces save different values, retain those observations for Microsoft support; PALO provides the evidence structure, not an answer about undocumented synchronization or Microsoft roadmap dates. You are seeing overlapping agent-access controls, but do not assume they represent independent policies. Microsoft currently documents Copilot > Settings > Data access > Agents for Copilot scenarios, while the centralized agent-settings documentation directs administrators to Agents > Settings > User access. Use the latter documented location for organization-wide access, and the Agent Registry for individual agent management. Copilot settings still cover broader Copilot-specific scenarios, including data access, billing, and product availability. Importantly, All users remains subject to existing app policies and user assignments; it is not blanket permission to use every agent. The documentation does not establish that both screens share identical backend configuration objects or guarantee a replacement date. Before relying on synchronization, compare the saved values and effective access using a small pilot group, following your change process. If the two views disagree, request tenant-specific confirmation from Microsoft support rather than assuming either overrides the other.
6:39 PM