<- All intelligenceThis Quarter
CISO
CIO
Compliance Officer
Enterprise Architect
Industry relevance
Financial Services
Healthcare
Government
SEPTEMBER 24, 2026
Microsoft Purview Network Data Security With Microsoft Entra Global Secure Access Reaches General Availability and Covers On-Behalf-Of Agent Traffic, September 2026
Network data loss controls now cover employees and agents acting on their behalf. Where agents with their own identity fall is a scoping decision someone has to document.
On September 24, 2026, Microsoft announced general availability of the integration between Microsoft Purview and Microsoft Entra Global Secure Access, which brings data security to the network. Microsoft states that context-aware Microsoft Purview classification and policies are enforced by Microsoft Entra at the network layer, across human actions and on-behalf-of agentic traffic. Microsoft gives the example of an employee or on-behalf-of agent trying to upload a sensitive document to an unsanctioned AI tool, where the policy can stop the transfer before the data leaves.
GOVERNANCE IMPLICATION
Data classification applied in Microsoft Purview is enforced by Microsoft Entra at the network layer, so one policy can follow a sensitive file toward an unsanctioned AI destination whether a person or an agent acting on that person's behalf starts the transfer. The announcement names those two categories. An agent that holds its own identity and acts without a user in the loop is a third category. Policy design needs an explicit decision on how that traffic is classified, logged, and blocked, recorded under a named owner. Evidence that a transfer was blocked will exist in product telemetry. Evidence that someone decided which agents the policy covers will exist only if the organization wrote it down.
SCENARIO
Hypothetical: An autonomous research agent with its own identity uploads a spreadsheet labeled confidential to a consumer AI tool. The network policy was written for employees and on-behalf-of agents. The investigation finds no policy decision that names autonomous agents, either to include or to exclude them.
THE GOVERNANCE QUESTION
The announcement names human and on-behalf-of agent traffic. Where does an autonomous agent with its own identity sit in your network data security policy, and who decided?
CONTROL GAP
Policy scope defined by traffic category, human or on-behalf-of agent, leaves autonomous agent traffic as an explicit design decision. Without a recorded owner for that decision, the control exists in the product and not in governance.
PRIMARY SOURCE
What's new in Microsoft Security: September 2026
Alym Rayani, Microsoft
September 24, 2026
Read the primary source ->(opens in new tab)CONTINUE READING
Read the next intelligence note.
MAY 7, 2026
Agent SecurityMicrosoft's Trust-and-Verify DLP Model for Copilot Has No Equivalent Check for Agent Actions, May 2026
Microsoft Digital's Copilot governance guide, published May 7, 2026 and updated June 8, 2026, describes a trust-and-verify model for employee data handling: employees apply sensitivity labels, and Purview DLP automatically checks that work through auto-labeling, quarantining, and escalation to content owners, legal, and security teams. The guide states this model catches roughly one percent of cases where labeling goes wrong. The verification described applies to whether data is correctly labeled and accessible, not to actions an AI agent takes using that data.
MARCH 24, 2026
AccountabilityAgent 365 GA governs on-behalf-of agents only — autonomous AI teammates with independent identities remain in Frontier preview beyond May 1
The May 1 GA version of Agent 365 governs agents operating on behalf of licensed human users. Agents with their own mailboxes, OneDrives, and independent identities map to the Agent Identity Authentication flow, which stays in Frontier preview. Frontier trial licenses were extended through December 2026, signaling that autonomous identity capability is still in development.