CVE-2026-62712 is a heap-based buffer overflow (CWE-122) in the Windows Win32K kernel component, with Microsoft's related advisories (ZDI-26-542/618/619/620/621) tying the flaw to user-mode printer driver (UMPD) graphics callbacks such as UMPDDrvBitBlt, UMPDDrvStretchBlt and UMPDDrvRealizeBrush. A local attacker with valid low-privileged credentials can trigger the overflow through crafted GDI/printer-driver operations, with no user interaction required. Successful exploitation elevates the attacker from a standard user to kernel/SYSTEM level, yielding high confidentiality, integrity and availability impact on the host. Every Windows 10 (1607, 1809, 21H2, 22H2), Windows 11 (23H2 through 26H1) and Windows Server (2012, 2016, 2019, 2022) installation on the listed builds is affected, which spans most of the supported Windows fleet. There is no public proof-of-concept, no CISA KEV listing, and a low EPSS of 0.4%, indicating no confirmed exploitation to date.
What to do: Apply the patch for CVE-2026-62712 from Microsoft's security update on all listed Windows 10, Windows 11 and Windows Server builds, prioritizing multi-user systems such as RDS hosts and print servers where untrusted users can execute code. Until patched, restrict interactive logon rights on servers to trusted users, since a kernel win32k overflow of this type has no reliable workaround. Verify remediation against the build numbers listed in Microsoft's advisory rather than assuming a single KB applies to every branch.
| Product | Affected |
|---|
| Microsoft Windows 10 | 1607, 1809, 21H2, 22H2 |
| Microsoft Windows 11 | 23H2, 24H2, 25H2, 26H1 |
| Microsoft Windows Server | 2012, 2016, 2019, 2022 |
Estimated exposure
mass≈1 billion+ Windows installations (every unpatched install of the listed Windows 10/11/Server builds is affected) — Public estimates put the active Windows 10/11 device base on the order of 1.4 billion, with Windows Server deployed in millions of instances, and the affected version list covers nearly all currently supported Windows SKUs.
Order-of-magnitude estimate by the model from install counts, market share and public scan data it knows; verify before quoting.