If your organization runs hybrid identity with Microsoft Entra, assume someone may now hold the email addresses of your administrators and details of your tenant configuration. Nothing in Microsoft’s account suggests passwords or tokens leaked, so the likely next move is social engineering aimed at your most privileged people. Microsoft says no patching is needed on your side, which makes the real work a matter of detection and staff readiness.

What happened

According to BleepingComputer, Microsoft told administrators that attackers exploited a vulnerability in the API behind Entra Connect Health, a cloud service for monitoring the health of on-premises identity infrastructure. The flaw was introduced on July 23, and Microsoft found it on November 9, 2026. A server-side fix went live on November 19, and the investigation closed on November 20.

Microsoft’s notification says the activity was unauthorized read-only access to tenant-level metadata, and that it found no unauthorized configuration changes. The notice describes an expired certificate that was no longer in use but still sat in Connect Health storage and was still returned by the API. Microsoft says passwords, password hashes, authentication tokens, valid cryptographic keys, on-premises server and Active Directory topology details and Azure subscription information were not exposed.

Microsoft has not said how many organizations were affected, who was behind it or when the attacks began. BleepingComputer reports that more than a dozen administrators described receiving the notice on Reddit, and some managed service providers received separate notices for multiple customer tenants. Microsoft told the outlet it notified customers through Azure Service Health.

Why it matters

Identity services sit at the center of almost every company’s security model, so even a metadata-only leak has value to an attacker. Administrator addresses and configuration details are reconnaissance material. They let a criminal craft believable lures that reference real tenant details, aimed at the people who can approve access or reset credentials. This suggests the damage here comes later, through phishing, not from the read access itself.

The timeline also deserves attention. The flaw existed for roughly three and a half months before discovery, and Microsoft has not said when exploitation started. Customers cannot judge their exposure window from the information given. The flaw lived in a cloud-side API, so no customer patch could have prevented it, and no customer scanner would have caught it. That is the structural lesson of hybrid identity: part of your attack surface is code you do not run.

The pattern fits a wider trend we have covered, in which attackers speed up intrusions and lean on trusted platforms, as in the FBI warning about the ARTEX framework. Polished, well-informed lures are cheap to produce, and stolen context makes them better.

What it means for you

Start by checking Azure Service Health for a notice, and do it for every tenant you manage. MSPs should expect separate notices per customer and should confirm that each one reached a person who will act on it.

  • Brief your admins. Warn anyone with privileged roles to treat unexpected messages about tenant changes, support cases or licensing as suspect, even when they cite accurate details.
  • Enforce phishing-resistant MFA. Microsoft recommends it for privileged accounts. Check that no admin is exempt and that no weaker fallback method remains.
  • Review logs. Examine Entra audit and sign-in logs for odd activity, and review your Connect Health configuration for anything you did not set.
  • Run Microsoft’s script. It shows the kinds of metadata that could have been exposed. It displays current values, not what an attacker actually read, so use it to gauge exposure and not as proof of access.

Also consider how administrator contact addresses are used. Where possible, avoid publicly guessable mailboxes for privileged roles, and ask Microsoft through your account team what logging you can obtain for the affected period. Budget time for the question of how much of your identity monitoring depends on cloud-side services you cannot inspect.

What to watch

  • Whether Microsoft discloses the number of affected tenants, the start of exploitation or the identity of the attackers.
  • Reports of phishing campaigns that cite real tenant details against administrators in coming weeks.
  • Any sign that the exposed metadata fed later account takeovers, which would change the severity of this incident.
  • Further detail on how an expired, unused certificate stayed retrievable, and whether Microsoft changes its data retention practices.

For more coverage of identity and intrusion trends, see our cybersecurity topic page.

Sources