Jones Web Designs & Development is now The Jones Foundry. What it means for clients

VMware vCenter zero-day exploited since September: patch now

Broadcom has patched a critical vCenter flaw that a suspected Chinese espionage group used for more than two months. Patching is step one. Checking for signs of a break-in is step two.

Broadcom has released updates for a critical VMware vCenter zero-day, tracked as CVE-2026-123890, that a suspected Chinese state-sponsored group has exploited since early September 2026. According to BleepingComputer, the flaw lets an unauthenticated attacker with network access bypass authentication and gain administrative control over the virtual machines vCenter manages. If your organization runs VMware, or relies on a provider that does, this one deserves attention today.

What the VMware vCenter zero-day allows

vCenter Server is a management platform that organizations use to run virtualized environments. The flaw sits in the VMware Directory Service. SecurityWeek reports it carries a CVSS score of 9.8, and that a remote attacker with no login could obtain an authenticated session with administrator privileges.

Broadcom fixed it on November 19 in advisory VMSA-2026-0009. The same advisory covers four other vulnerabilities affecting VMware SDDC Manager and VMware Tools for Windows. Broadcom treats the vCenter fix as an emergency update and says there are no official workarounds. It does say network restrictions and firewall controls can help reduce exposure. BleepingComputer adds that the flaw affects vCenter whether or not Enhanced Linked Mode is turned on.

One detail worth noting: according to SecurityWeek, Broadcom's advisory does not mention exploitation in the wild. The exploitation reports come from Google's Mandiant, which reported the bug to Broadcom.

Who is affected and which versions are fixed

Broadcom says you are affected if you run any version of vCenter, SDDC Manager, or VMware Tools for Windows older than the fixed versions listed in the advisory. That includes standalone deployments and those bundled in VMware Cloud Foundation, VMware vSphere Foundation, and the Telco Cloud products.

SecurityWeek lists the vCenter fixes as versions 9.1.1.0200, 9.0.2.0200, and 8.0 U3m. Mandiant says the attackers' tool works against vCenter 7.x and 8.x. There is no public patch for version 7.0, which reached end of general support in October 2025. If you are still on 7.0, you are exposed to this flaw with no vendor fix available, so network restrictions and an upgrade plan matter even more.

How the attackers got in and moved around

Mandiant and Google Threat Intelligence Group attribute the attacks to a group they call Armory Castle (formerly UNC5221, also known as Warp Panda). The group is known for exploiting zero-days in enterprise network appliances, including Citrix NetScaler and Ivanti Connect Secure.

According to Mandiant, this campaign started with a different zero-day, CVE-2026-102764, in BeyondTrust Remote Support. From there the attackers moved into internal Linux systems, installed web shells for persistent access, escalated to root, and scanned the network for VMware systems. They then used a custom tool called BONESAW to exploit vCenter, create a single sign-on administrator account, and steal the server's directory database.

That is where it gets serious. BleepingComputer reports BONESAW can also extract SAML signing keys. A second tool, SAWSNAKE, uses them for so-called Golden SAML attacks that forge authentication tokens, bypass multifactor authentication, and impersonate real users. The attackers also deployed additional backdoors, reached ESXi hosts, and pulled sensitive data from virtual machines, including Windows Active Directory database files.

SecurityWeek adds that Mandiant has evidence the group knew about this flaw before a separate vCenter authentication bypass (CVE-2026-59309) was disclosed on July 29, and that BONESAW was used against a server already patched for that earlier bug. In other words, being current on the previous patch did not protect that victim.

Why patching alone may not be enough

The patch closes the door, but it does not remove anyone who already walked through. Mandiant CTO Charles Carmakal put it plainly, as quoted by SecurityWeek: "Organizations should both apply the patches and examine their systems for evidence of compromise."

Because the attackers can steal signing material and credentials, a patched server may still have a rogue administrator account or forged tokens in circulation. Treat this as a possible incident, not just a maintenance task.

What to do this week

If you manage VMware yourself, or you pay someone who does, work through this list:

  • Ask your IT team or provider whether you run vCenter, SDDC Manager, or VMware Tools for Windows, and which versions.
  • Apply the fixed versions from Broadcom's advisory VMSA-2026-0009 as soon as you can.
  • Limit network access to vCenter and other management interfaces with firewall rules, so they are not reachable from the internet or from systems that do not need them.
  • Review administrator accounts, especially any new single sign-on admin accounts you do not recognize, and review logs for unusual activity going back to early September.
  • Check any BeyondTrust Remote Support appliances are patched, since that was the reported way in.
  • If you find signs of compromise, plan to rotate credentials and bring in incident response help rather than relying on the patch.
  • If you are on vCenter 7.0, make an upgrade plan and tighten network restrictions in the meantime.

What this means if you do not run VMware

Many small organizations never touch vCenter directly, but a hosting company, managed IT provider, or agency might. It is reasonable to ask them in writing whether they are affected and what they have done. The broader lesson also applies to everyone: remote support tools and management consoles are high-value targets, so keep them patched and off the open internet where possible.

Want to know how exposed your own site is? Start with a free external scan.

Sources

This post was drafted with AI from the reporting linked above and published by The Jones Foundry. For full details, read the original sources.

Found this useful? Pass it on to someone who would want to know.

All Online Security posts →