Attackers exploited a flaw in the Microsoft Entra Connect Health API to query administrators’ email addresses and other tenant metadata, Microsoft confirmed. The company fixed the issue on its servers on November 19 and notified affected organizations; it found no unauthorized configuration changes.
In a notification sent to administrators, Microsoft said its investigation concluded on November 20 and confirmed read-only access to metadata associated with organizations. The vulnerability could have allowed Connect Health monitoring configurations to be modified, but Microsoft said it found no evidence that anyone had done so.
“We identified and addressed a security issue affecting Microsoft Entra Connect Health that could have resulted in unauthorized access to tenant-level contact and configuration metadata. We deployed a fix on November 19, 2026, and have notified impacted customers through Azure Service Health, providing guidance and support”
— Microsoft spokesperson
The statement was sent to BleepingComputer, which published the news on November 21. Microsoft did not provide the outlet with further details about the breach. The company has also not identified those responsible, specified when the attacks began, or reported how many tenants were exposed.
What data was accessible
Connect Health is a cloud service that lets users monitor the health, performance, and alerts of on-premises identity infrastructure from the Microsoft Entra admin center. The affected API contained tenant and service monitoring metadata, including information that could provide context for an impersonation attempt targeting an organization’s administrators.
| Accessible through the flaw | Microsoft said was not exposed |
|---|---|
| Email addresses of technical contacts and administrator accounts | Passwords and password hashes |
| Tenant IDs, display names, initial domains, country codes, and regional information | Authentication tokens and valid cryptographic keys |
| Entra license tiers and directory synchronization status, including whether it was enabled and the date of the last sync | Details of on-premises server topology and Active Directory |
| Connect Health onboarding status | Azure subscription information |
| An expired certificate that had been used to register Connect Health agents |
The expired certificate had been used to register Connect Health agents with an internal Microsoft service that was retired in late 2024. According to the company, it was no longer in use, but remained stored in Connect Health and was still being returned by the API. Microsoft clarified that no valid cryptographic keys were exposed.
The exposure did not include passwords, password hashes, authentication tokens, or valid cryptographic keys. However, combining identified contacts with specific tenant data can make fraudulent messages seem credible. In his analysis of the incident, Enzo Soler warns of the risk of phishing emails or calls targeting administrators with real details about their organization.
The timeline: from July to notifications
According to Microsoft’s notification, as reported by BleepingComputer, the vulnerable capability was introduced on July 23, 2026. Microsoft detected the issue on November 9. It then reviewed available telemetry to identify unauthorized API calls affecting customers.
On November 19, it deployed a server-side fix to prevent further exploitation. The following day, it concluded its investigation and began notifying affected organizations. The company said no software updates or other client-side remediation steps are required to fix the vulnerability.
Microsoft has not published the number of affected organizations. BleepingComputer noted that more than a dozen administrators commented in a Reddit post that they had received the notification; some managed service providers also said they had been contacted by several customer tenants. One administrator said the messages arrived hours apart. These accounts do not establish the full scope of the incident.
Soler added that, at the time of his publication, there was no CVE or public security advisory. He also described the service as part of hybrid environments that may include Entra Connect synchronization servers, AD FS, and domain controllers. BleepingComputer, meanwhile, explains that Connect Health monitors on-premises identity infrastructure from the Entra admin center.
What Microsoft recommends organizations do
Microsoft asks organizations to alert their security teams so they can watch for potential phishing and social engineering campaigns targeting the exposed administrative email addresses. It also recommends phishing-resistant multifactor authentication for privileged accounts, reviewing Connect Health configurations, and checking Entra audit and sign-in logs for suspicious activity.
The script is published at aka.ms/ConnectHealth/scripts and lets users view the types of tenant metadata. It has an important limitation: it shows current configuration values, not the specific data accessed during the incident. Therefore, it cannot reconstruct what information an attacker read.
Soler specifies that FIDO2 security keys and passkeys count as phishing-resistant methods, while SMS codes and push approvals do not. He also advises informing identified contacts that they may receive messages mentioning the tenant ID or license tier: those details alone do not prove that an email is legitimate.
The scope remains unknown
The fix has been deployed, and Microsoft says no installations or software changes are needed to close the vulnerability. The remaining task for each organization is to review its own logs and configurations and prepare exposed contacts for impersonation attempts. Microsoft’s script can show the current state, but not which metadata was read during the unauthorized access.
Microsoft has not said who exploited the vulnerability or how many tenants were affected. It has also not specified when the illegitimate queries began. The company’s investigation ended on November 20; its customer notification and security recommendations are, for now, the information it has shared about the incident.
Sources: BleepingComputer, Enzo Soler