Attackers Exploit Microsoft Entra Connect Health Flaw To Access Administrator Emails and ...
Microsoft has confirmed that attackers exploited a flaw in Entra Connect Health, its cloud service for monitoring identity infrastructure, to read customer records containing administrator email addresses and details of how affected organisations manage their directories. Microsoft fixed the vulnerability on 19 November and notified affected customers through Azure Service Health. The company has not disclosed how many tenants were affected or who was responsible.
For security teams, the most consequential detail is what the exposed records linked together. Someone reading an affected tenant’s record could identify a technical or administrative , see which Microsoft Entra licence tier the organisation used and determine whether its directory was synchronising with the cloud. Microsoft has warned customers to watch for targeted phishing that draws on those details. Its investigation found unauthorised reads and no evidence that attackers changed monitoring settings.
Nearly Four Months Between Introduction and Fix
According to a copy of Microsoft’s customer notification shared by an administrator , the flaw entered the Connect Health application programming interface on 23 July 2026 . Microsoft discovered it on 9 November and deployed a service-side correction ten days later. On 20 November , investigators concluded that an unauthorised party had read metadata belonging to the organisation that received the notice.
Microsoft said it examined service telemetry to identify unauthorised API calls and determine which customers to . The dates establish how long the flaw was present. The first confirmed malicious request remains undated in the public account, leaving open whether attackers discovered the weakness early in that period or shortly before Microsoft did.
How a Monitoring Service Came to Hold Identity Data
Organisations that maintain an on-premises Active Directory alongside Microsoft Entra ID need to know when the two environments stop synchronising or when a federation service develops a fault. Entra Connect Health gives administrators a central view of those systems. Its monitoring covers directory synchronisation, Active Directory Federation Services and Active Directory Domain Services, according to Microsoft’s product documentation . Microsoft Learn
That role explains both the contents of the exposed records and the limits of the incident. A last-sync timestamp can tell an administrator when a directory last communicated with the cloud. It can reveal something the organisation’s infrastructure to an outsider, too. Microsoft said the vulnerable API returned that kind of operational information, while passwords, password hashes, authentication tokens and valid cryptographic keys remained outside the exposure. It also ruled out exposure of Azure subscription information and details of on-premises servers or Active Directory topology.
The distinction also matters for patching. Microsoft corrected the API on its own infrastructure. Customers do not need to identify an affected Entra Connect build or install an update on their synchronisation servers to close this vulnerability.
What the Accessed Records Revealed
Microsoft’s notification described the fields present in the API responses accessed for the recipient’s tenant. Alongside email addresses for technical contacts and administrator accounts, the records carried the organisation’s tenant identifier, display name, original domain name, country code and regional scope. They also showed its Entra licence tier, whether directory synchronisation was enabled, the time of the last synchronisation and its Connect Health onboarding status.
One field had outlived the system it served. Connect Health had previously used a certificate to register monitoring agents with an internal Microsoft service. Microsoft retired that service in late 2024, but the expired certificate remained in storage and the API continued returning it. Microsoft told customers the certificate was no longer usable for that purpose. Its presence explains why a certificate appears in the breach notification even though investigators ruled out exposure of valid keys.
Investigators also considered whether anyone had altered a customer’s monitoring configuration. Microsoft said the flaw could have permitted such changes. The activity it identified was read-only, a finding that narrows the confirmed impact to disclosure of information.
Why Microsoft Is Warning Administrators Phishing
An administrator receiving a message a synchronisation problem might reasonably expect a genuine support to know the organisation’s domain or recognise that it runs a hybrid directory. After this incident, those details carry less weight as proof of who sent a message. The exposed information could help someone address the right recipient and a service the organisation actually uses. Microsoft identified this as a follow-on risk. Its notification did not report an ensuing phishing campaign or account takeover.
Microsoft asked customers to brief global administrators, security operations staff, IT teams and the people listed as tenant technical contacts. It also advised checking that those details are current. For the notification itself, customers can independently open the Azure portal, navigate to Service Health → Security advisories and verify that their tenant is listed. Managed services provider Magna5 identified the notice as YZRN-MK0 in its customer advisory . Microsoft’s Azure Service Health documentation explains which permissions are needed to see sensitive incident details.
For privileged accounts, Microsoft recommended phishing-resistant multifactor authentication and, where applicable, time-limited activation of administrative roles through Privileged Identity Management. One practical question for identity teams is what an administrator’s sign-in policy actually requires. Entra supports a specific phishing-resistant authentication strength that can require methods such as FIDO2 security keys, Windows Hello for Business or multifactor certificate-based authentication. A general MFA requirement accepts a wider range of methods. Microsoft’s authentication strength guidance sets out that distinction.
What Customers Can Establish From Their Own Records
Microsoft has provided a read-only PowerShell script that administrators can use to inspect the tenant fields involved. It answers a specific question what those fields contain today . If an organisation changed its technical after the incident, for example, the script would show the replacement address. It cannot establish which address appeared in the response an attacker received months earlier.
Customer logs serve a different purpose. Microsoft advised organisations to compare their current Connect Health settings with their expected configuration and review Entra sign-in logs, Entra audit logs and Azure Activity Logs for unusual administrative activity. Those checks can help investigate a suspicious sign-in or an unexplained settings change. Microsoft relied on its own service telemetry to identify the unauthorised API reads behind the customer notices.
There is a practical limit on how far a customer can look back. Under Microsoft’s Entra retention policy , standard audit and sign-in records are retained for seven days on Entra ID Free and 30 days on P1 and P2. Both periods are shorter than the interval between the flaw’s introduction in July and its repair in November. An organisation that had already routed those logs to longer-term storage may therefore have a substantially fuller record of activity around its privileged accounts.
The Scale of the Incident Remains Undisclosed
Several administrators publicly reported receiving the warning, and some managed service providers said notices arrived for more than one customer tenant. Those accounts establish that the disclosure reached multiple organisations, but they offer little basis for estimating its overall scale. As of 22 November 2026 , Microsoft had not published an affected-tenant count, attributed the activity to a threat actor or said when exploitation began. For an individual organisation, the most specific evidence remains its own Azure Service Health advisory, which ties Microsoft’s investigation to the tenant concerned.
