TRENDING
A person sweeping a metal detector across a wet sandy beach with a sea fort on the horizon
November 22, 2026
How to Write YARA Rules With YARA-X to Hunt Malicious Python .pth Startup Files and Measure False Positives Before You Trust Them
A peeling black-and-white metal door standing ajar, with a dark empty hall and bright windows beyond
November 22, 2026
AWS’s “Found a Zero-Day on Their Own” Post Never Names the Product; the Advisory Behind Its Link Shows a Server That Ships With No Login, a README Now Saying Keep It Off the Internet, and a Fix Merged October 19 That PyPI and Docker Hub Still Do Not Carry
A hand paging through index cards in an open wooden library card-catalog drawer labeled Agriculture, with a second drawer open below to the right
November 22, 2026
Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts
Swimmers in separate lanes of a blue swimming pool, seen from above the water, with lane lines running across the picture
November 22, 2026
How to Size a PostgreSQL Connection Pool in Python: Find Where Throughput Stops Growing, Split max_connections Across Workers, and Stop Slow Handlers From Hoarding Connections
Workers in caps lean over a 1913 Ford assembly line, each assembling parts as they pass on the moving line
November 22, 2026
OpenAI’s Performance Factory Turns Regression Response Into a Task Queue for Agents, but Its Numbers Stop at 600 Root-Caused Cases in 1,700 Investigations and a 70% Grade on 100 Suggestions, and in a Toy World Each of Its Reviewer’s Two Counterfactual Questions Can Return a Confident Wrong Answer
22 Nov 2026
SXZ.io SXZ.io
  • Home
Search the Site
Popular Searches:
Technology Amazon AI
Recent Posts
The sun setting over the sea behind a pyramid-shaped stone monument at Cape Soya, the northernmost point of Japan
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
November 22, 2026
Brass sextant in an open wooden case inside a museum display, an instrument for measuring the angle to the horizon
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
November 22, 2026
Brass telescope on a wooden tripod inside a white observatory dome with the roof slit open to the sky
Rackspace Says Leaked Credentials Let an Intruder Act as Legitimate Users in Some Spot Customer Organizations and Reach Their Kubernetes Secrets, a Platform9 Co-Founder Names a Breached Observability System as the Root Cause, and the Status Page Says Only “System Maintenance Operations”
November 22, 2026
SXZ.io SXZ.io
  • Home

Categories

Articles 310 Posts
News 316 Posts
Learning Hub 282 Posts
Home/News/Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts
News

Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts

Microsoft told customers that unauthorized parties read tenant-level metadata, including administrator and technical-contact email addresses, through an Entra Connect Health flaw fixed November 19;...

November 21, 2026 10 Min Read
21

Microsoft has told customers that unauthorized parties used a flaw in the Microsoft Entra Connect Health API to read tenant-level metadata, including the email addresses of technical contacts and administrator accounts. BleepingComputer reported the notice on November 21. Microsoft says it deployed a server-side fix on November 19 and that no customer action is required to address the flaw.

Table Of Content

  • What Microsoft told customers
  • What Microsoft’s reference scripts show
  • The tenant object holds every field category in the notice
  • The service-level objects hold the details the notice says stayed private
  • The scripts are read-only and signed
  • A timeline from public records
  • What nobody has published yet
  • Why the email list matters
  • What administrators can check today
  • Limits of this report

We could not read the notice itself. It was shared in a Reddit thread that blocks scripted readers, and the one archived copy we found holds only an empty page shell, so every notice sentence below is as BleepingComputer quotes it. What we could read first-hand is Microsoft’s own reference-script repository for Connect Health, which shows what the tenant object holds, and the Microsoft Learn pages that say who receives Microsoft’s tenant-level security notices. Both change how the notice reads.

What Microsoft told customers

BleepingComputer describes Connect Health as “a cloud-based monitoring service that allows administrators to monitor the health, performance, and alerts of their on-premises identity infrastructure through the Microsoft Entra admin center.” Per its report, Microsoft found the flaw in the Connect Health API on November 9, “after it was first introduced on July 23.” The quoted text does not say which check failed.

The public wording differs by speaker, and the differences are worth reading closely.

Statement Speaker What it describes
“unauthorized read-only access to tenant-level metadata associated with your organization had occurred” Customer notice, as quoted by BleepingComputer What happened to the recipient’s tenant
“We did not identify any unauthorized configuration changes” Customer notice, as quoted What was not found
“allowing unauthorized parties to access tenant and service-monitoring metadata and potentially modify Connect Health monitoring configurations” BleepingComputer’s summary of the investigation What the flaw allowed
“could have resulted in unauthorized access to tenant-level contact and configuration metadata” Microsoft spokesperson to BleepingComputer The public statement

Read together, the first two rows are findings for a specific tenant. The third is a capability that includes “service-monitoring metadata,” which the notice sentence does not mention. The fourth is softer than the first: “could have resulted” against “had occurred.” The two can be consistent, because BleepingComputer says Microsoft notified impacted customers, but a reader of the spokesperson’s sentence alone would not learn that Microsoft had confirmed access for those customers. We cannot resolve the difference without the notice text.

Per BleepingComputer, Microsoft also says the flaw “did not expose user passwords, password hashes, authentication tokens, valid cryptographic keys, on-premises server and Active Directory topology details, or Azure subscription information.” BleepingComputer adds that Microsoft “reviewed available telemetry to identify unauthorized API calls affecting customers.”

What Microsoft’s reference scripts show

BleepingComputer says Microsoft also published a PowerShell script at aka.ms/ConnectHealth/scripts. That link answers with a 301 redirect to the ConnectHealth folder of the AzureAD/EntraIDGovernance repository on GitHub: nine PowerShell scripts, a README, a SampleOutputs.md file and a one-byte placeholder file. The README calls them “read-only PowerShell scripts” that are “provided as references so you can inspect the information returned for your tenant by the Microsoft Entra Connect Health APIs.”

The tenant object holds every field category in the notice

The sample output for GetTenant.ps1 is a tenant configuration object with 26 fields. The table maps the notice’s four bullets, as BleepingComputer quotes them, to the fields that match. The mapping is ours, because the notice names no fields, and the sample values are illustrative (Microsoft says so in the file).

Notice bullet (BleepingComputer’s quotation) Matching fields in the sample object
“Email addresses belonging to tenant technical contacts and administrator accounts” technicalNotificationEmails, globalAdminsEmail
“Tenant IDs, display names, initial domain names, country codes, and regional information” tenantId, tenantName, initialDomain, countryLetterCode, replicationScope
“Microsoft Entra license tiers and directory synchronization status, including whether synchronization was enabled and the last synchronization timestamp” aadLicense, aadPremium, isDirSyncEnabled, lastDirSyncTime
“An expired certificate previously used to register Connect Health agents with a Microsoft internal service that was decommissioned in late 2024” pksCertificate

Twelve of the 26 fields match the notice’s list. The other 14, among them createdDate, agentAutoUpdate, alertSuppressionTimeInMins and tenantServices, are not in the list, and nothing we could read says whether they were readable. The script below, which anyone can run, reproduces the count from the sample file as it stands in the November 20 revision of the repository.

# tenant_fields.py
# Reads Microsoft's GetTenant sample output from the Connect Health reference-script repository
# (pinned to the November 20 revision), lists its fields and groups them by the four categories
# in the customer notice. The grouping is ours; the notice names no fields.
import json
import re
import urllib.request

COMMIT = "c006d434242c67c7d6d5c428f6be0be51b4003f7"
URL = ("https://raw.githubusercontent.com/AzureAD/EntraIDGovernance/"
       + COMMIT + "/ConnectHealth/SampleOutputs.md")

text = urllib.request.urlopen(URL, timeout=60).read().decode("utf-8")
section = text.split("## GetTenant.ps1")[1].split("## GetServices.ps1")[0]
sample = json.loads(re.search(r"```json\n(.*?)```", section, re.S).group(1))

groups = {
    "admin and contact email lists": ["globalAdminsEmail", "technicalNotificationEmails"],
    "tenant identity and region": ["tenantId", "tenantName", "initialDomain",
                                   "countryLetterCode", "replicationScope"],
    "license and sync status": ["aadLicense", "aadPremium", "isDirSyncEnabled", "lastDirSyncTime"],
    "certificate": ["pksCertificate"],
}
mapped = {name for names in groups.values() for name in names}

print("fields in the sample:", len(sample))
for label, names in groups.items():
    present = [n for n in names if n in sample]
    print(f"{label}: {len(present)} of {len(names)} present -> {', '.join(present)}")
rest = [name for name in sample if name not in mapped]
print("fields the notice's list does not cover:", len(rest))
print(", ".join(rest))

It prints:

fields in the sample: 26
admin and contact email lists: 2 of 2 present -> globalAdminsEmail, technicalNotificationEmails
tenant identity and region: 5 of 5 present -> tenantId, tenantName, initialDomain, countryLetterCode, replicationScope
license and sync status: 4 of 4 present -> aadLicense, aadPremium, isDirSyncEnabled, lastDirSyncTime
certificate: 1 of 1 present -> pksCertificate
fields the notice's list does not cover: 14
devOpsTtl, createdDate, disabled, lastDisabled, tenantAdoptionJobLastRunDateTime, lastVerified, onboardingAllowed, onboarded, consentedToMicrosoftDevOps, alertSuppressionTimeInMins, agentAutoUpdate, disabledReason, tenantInQuarantine, tenantServices

The email fields are the part that matters most. In the sample, globalAdminsEmail is a list (four made-up addresses) next to technicalNotificationEmails. That fits the notice’s “administrator accounts” and “technical contacts,” and it suggests the exposed list covers a tenant’s Global Administrators rather than a single contact.

The service-level objects hold the details the notice says stayed private

The sample objects for the other scripts carry fields such as machineName, osName, osVersion and installedQfes for each monitored server, and forestName, domains and sites for an Active Directory forest. Those fit the category the notice says was not exposed, “on-premises server and Active Directory topology details” (our reading). The line between the tenant object and these service objects is where the open question sits: BleepingComputer’s summary says the flaw allowed access to “service-monitoring metadata,” while the notice sentence it quotes limits what happened to “tenant-level metadata.”

The scripts are read-only and signed

Eight of the nine scripts send HTTP GET requests to Azure Resource Manager under https://management.azure.com/providers/Microsoft.ADHybridHealthService/, always with api-version=2014-01-01: the tenant configuration, the service list and each service, its members, its monitoring configurations and, for Active Directory Domain Services registrations, the forest and replication summaries. The ninth only signs in. We found no write cmdlets and no method other than GET in any of them. Microsoft has not said which endpoint the flaw affected.

All nine scripts carry a valid Authenticode signature from Microsoft Corporation in Redmond. The check below ran on Windows PowerShell 5.1 against the files from the November 20 revision. Paste it at a prompt in the folder that holds the scripts; saved as a file in that folder, it would list itself too. With phishing against administrators expected, it is a cheap habit before running anything from a link, however official the link looks.

# Run in Windows PowerShell 5.1 or PowerShell 7 on Windows, in the folder that holds the downloaded scripts.
Get-ChildItem .\*.ps1 | ForEach-Object {
    $sig = Get-AuthenticodeSignature -FilePath $_.FullName
    '{0} | {1} | {2}' -f $_.Name, $sig.Status, $sig.SignerCertificate.Subject
}

It printed:

GetForestSummary.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetReplicationSummary.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetService.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetServiceMembers.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetServiceMonitoringConfigurations.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetServices.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetTenant.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
GetTenantPksCertificateDetails.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US
SelectConnectHealthTenant.ps1 | Valid | CN=Microsoft Corporation, O=Microsoft Corporation, L=Redmond, S=Washington, C=US

The signature check is our suggestion, not Microsoft’s. BleepingComputer’s caveat is that the script “displays current configuration values rather than the information that was accessed during the breach,” which matches the README’s description of what the scripts return for a tenant. The certificate script, GetTenantPksCertificateDetails.ps1, decodes the tenant’s pksCertificate value and prints, among other fields, NotAfterUtc and HasPrivateKey, so an administrator whose tenant has a populated value can see whether the stored certificate has expired and whether it carries a private key. The field is our best match for the expired certificate in the notice, not a confirmed one.

The history of the folder adds a small timeline. It was created on November 16 with a placeholder file, the scripts first went in on November 18 without signatures, and signed copies followed five minutes later. All nine were uploaded again on November 20 at 16:43 UTC, the day customers were notified, with an optional -TenantId parameter (validated as a GUID) and a version header. The parameter fits work across several tenants, which matches the reports of managed service providers receiving several notices, but Microsoft does not say that is why.

A timeline from public records

Date (UTC) Event Source
July 23 Vulnerability first introduced Notice, as relayed by BleepingComputer
November 9 (Monday) Microsoft discovers it Same
November 16, 19:25 Script folder created on GitHub with a placeholder file GitHub commit history
November 18, 15:36 Scripts, README and sample outputs uploaded without signatures; signed copies at 15:41 GitHub commit history
November 19 (Thursday) Server-side fix deployed Microsoft spokesperson
November 20 (Friday) Investigation concluded and customers notified; scripts re-uploaded at 16:43 Notice; GitHub commit history
November 21, about 18:00 BleepingComputer reports the notice BleepingComputer

By our arithmetic, that is 109 days from introduction to discovery, 10 days from discovery to the fix, 119 days from introduction to the fix and 11 days from discovery to notification.

What nobody has published yet

  • How many tenants. BleepingComputer says Microsoft has not disclosed how many organizations were affected. “More than a dozen administrators” reported the notice in the Reddit thread, and some managed service providers reported notices for several customer tenants.
  • Who and when. The same report says Microsoft has not disclosed who exploited the flaw or when the attacks first began. The flaw is dated July 23 and was fixed November 19, a window of 119 days. BleepingComputer says Microsoft reviewed “available telemetry”; how far back that telemetry reaches is not stated.
  • A CVE. At about 19:20 UTC on November 21, an NVD keyword search for Entra Connect Health returned no CVE records, the CISA Known Exploited Vulnerabilities catalog (version 2026.11.20) had no Connect Health entry, and Microsoft’s November CVRF (Common Vulnerability Reporting Framework) document (2,264 vulnerability entries, released at 01:22 UTC that day) had no entry that mentions Connect Health. Microsoft has not said whether it will assign one.
  • The mechanism. Nothing we read says which check failed in the API or whether the callers needed any credentials. We did not try to reproduce anything, and this post does not speculate.

Why the email list matters

Microsoft Learn says tenant-level security notices from Azure go to the same kinds of people the notice lists as exposed. Its security notifications page states: “We send notifications to Global Admins, Technical Contacts, and Security Admins,” and adds that “For services designed at the tenant level (for example, Microsoft Entra), notifications are sent at the tenant level.” The notice says the addresses of technical contacts and administrator accounts were readable. Our reading is that those include the inboxes where a real follow-up from Microsoft would land, which is also where a look-alike message would aim. Per BleepingComputer, Microsoft recommends that organizations alert their security teams to potential targeted phishing and social-engineering attacks.

Learn also says who gets a breach notice. The tenant Properties page describes the Global privacy contact as “also who Microsoft contacts if there’s a data breach related to Microsoft Entra services,” and adds that if none is listed, “Microsoft contacts your Global Administrators.”

What administrators can check today

These steps are ours, drawn from Microsoft’s notice as BleepingComputer relays it and from the Learn pages linked here. Microsoft says no software update or other customer remediation is required.

  1. Read the notice in the channel Microsoft names. The spokesperson says customers were notified through Azure Service Health. Treat an email that asks you to click through to confirm the notice, or to run a script from a link, as suspect until it matches what Service Health shows.
  2. Check who is on the receiving end. In the Microsoft Entra admin center, open Entra ID, Overview, Properties and confirm that the Technical contact and the Global privacy contact are monitored mailboxes. In Connect Health, open Sync errors and Notification settings to see whether all global administrators or a custom list receive alerts (Learn: Connect Health operations).
  3. Require phishing-resistant sign-in for privileged roles. The built-in Phishing-resistant MFA authentication strength is a Conditional Access control. It allows Windows Hello for Business or a platform credential, a FIDO2 security key, or Microsoft Entra certificate-based authentication (multifactor), as listed on the authentication strengths page.
  4. Know how far back your logs go. By default, Entra audit and sign-in logs are kept seven days on the Free edition and 30 days on P1 and P2 (Learn: data retention), so looking across a 119-day window needs logs that were already exported; Learn describes routing them to an Azure storage account with Azure Monitor. Microsoft’s notice advises reviewing them, and Microsoft has not said whether the unauthorized calls left any record in a customer’s own tenant logs. The same retention limit shaped our look at the FBI’s ARTEX warning.
  5. Optionally look at what your tenant returns. Verify the script signatures first, run the scripts only against your own tenant, and remember they show current values, not an access record.

Limits of this report

We did not run Microsoft’s scripts against a tenant, did not call any Connect Health or Azure Resource Manager endpoint, and did not find the notice text outside BleepingComputer’s quotations. The field mapping and the reading of the email lists are ours and are labeled as such. If Microsoft publishes the notice, a CVE or a count of affected tenants, those facts should replace the gaps listed above. For related Entra tenant incidents we have covered, see the ConnectWise cloud backup compromise.

Tags:

Cloud SecurityData BreachesEntra Connect HealthIdentity ManagementMicrosoftMicrosoft Entra IDPhishing

Share

Swimmers in separate lanes of a blue swimming pool, seen from above the water, with lane lines running across the picture
Previous Post

How to Size a PostgreSQL Connection Pool in Python: Find Where Throughput Stops Growing, Split max_connections Across Workers, and Stop Slow Handlers From Hoarding Connections

A peeling black-and-white metal door standing ajar, with a dark empty hall and bright windows beyond
Next Post

AWS’s “Found a Zero-Day on Their Own” Post Never Names the Product; the Advisory Behind Its Link Shows a Server That Ships With No Login, a README Now Saying Keep It Off the Internet, and a Fix Merged October 19 That PyPI and Docker Hub Still Do Not Carry

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Latest
22 Nov
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
22 Nov
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
Trending
November 22, 2026
How to Retire an API Endpoint in FastAPI: Send Deprecation and Sunset Headers, Find the Clients Still Calling It, Schedule Brownouts, and Block an Early Removal With oasdiff
November 22, 2026
MIT Technology Review’s “Most Misunderstood Graph” Update Reports METR’s Judgment Findings Mostly Fairly, but It Never Says Each Point Is a 50% Success Rate or Gives the New 3.6-Month Doubling Time and 2.1 to 4.9 Month Range, and It Credits OpenAI With a Comparison METR Built
November 22, 2026
Rackspace Says Leaked Credentials Let an Intruder Act as Legitimate Users in Some Spot Customer Organizations and Reach Their Kubernetes Secrets, a Platform9 Co-Founder Names a Breached Observability System as the Root Cause, and the Status Page Says Only “System Maintenance Operations”
November 21, 2026
How to Write YARA Rules With YARA-X to Hunt Malicious Python .pth Startup Files and Measure False Positives Before You Trust Them
November 21, 2026
AWS’s “Found a Zero-Day on Their Own” Post Never Names the Product; the Advisory Behind Its Link Shows a Server That Ships With No Login, a README Now Saying Keep It Off the Internet, and a Fix Merged October 19 That PyPI and Docker Hub Still Do Not Carry
November 21, 2026
Microsoft Tells Customers Unauthorized Parties Read Admin Email Addresses and Tenant Metadata Through an Entra Connect Health Flaw Introduced July 23, but Gives No Tenant Count or Start Date, and Its Own Docs Send Tenant Security Notices to the Same Kinds of Contacts

Related Posts

Rows of server racks in a data center representing network infrastructure targeted by botnets
News

C0XMO Botnet Shows Why Old Router Firmware Still Matters

June 7, 2026
Close-up of a USB flash drive, representing physical data-theft risk in office security incidents
News

Fake IT Support Is Now Walking Through the Front Door

June 7, 2026
A phone security app on a smartphone resting on a laptop keyboard.
News

Everest Forms Pro Flaw Is Being Exploited to Create Rogue WordPress Admins

June 7, 2026
A phone secured by a padlock, illustrating AI data-leak containment and security controls.
News

OpenAI’s Lockdown Mode Is a Data-Leak Brake, Not a Prompt-Injection Cure

June 8, 2026
SXZ.io SXZ.io
  • [email protected]

Categories

Articles
Learning Hub
News

All Rights Reserved by SXZ.io ©2026