The CVSS score of 7.8 for CVE-2026-62772 materially understates the risk because this vulnerability exists in a container isolation filter driver — a component whose failure mode is not merely local privilege escalation but potential container-host boundary collapse. Filter drivers operating in kernel mode sit at the intersection of untrusted container I/O requests and host filesystem operations. A heap overflow in unionfs.sys doesn't just let a container workload become root on its own VM — it could let that workload escape to the host kernel. The CVSS metrics treat this as a standard local EOP, but container isolation drivers are trusted components whose compromise invalidates the fundamental security boundary they exist to enforce. The 'authorized attacker' language in the description is particularly dangerous for container threat models: authorization is enforced BY the isolation boundary we're worried about bypassing. Treating a container workload as 'authorized' to exploit a filter driver flaw is like treating a process as 'authorized' to escape its sandbox — it assumes the very integrity the vulnerability questions. The 26H1 designation is also concerning: this is a forward-looking release branch, meaning the vulnerability code is entering deployment pipelines now rather than being retired from legacy systems. This code will likely be backported into mainstream Windows releases and LTS container host images without re-review, compounding its persistence. Microsoft has published seventeen advisories between 2015 and 2024 addressing heap management errors in filesystem filter drivers — the same pattern, the same root cause — yet the identical flaw appears in container isolation drivers. This suggests institutional failure rather than impossible engineering requirements. HVCI addresses exploitation technique, not root cause: a heap overflow that can't be reliably exploited via known HVCI techniques still exists in the codebase. Assess whether Azure Container Instances, Windows Server Containers, and WSL2 are affected. The question isn't how severe this CVE is on its own terms — it's what percentage of your container fleet depends on this filter driver, because CVSS treats vulnerabilities as independent events while container infrastructure treats filter drivers as load-bearing walls.
CVE-2026-62772
Official description Straight from the sourceThe vendor's or NVD's own wording, published unedited. Authoritative, but often terse — it says what broke, rarely what to do.
NVD · uneditedHeap-based buffer overflow in Windows Container Isolation FS Filter Driver (unionfs.sys) allows an authorized attacker to elevate privileges locally.
In the news
Third-party coverageSurfaced from public web coverage — external links open in a new tab.
Technical summary Written by usOur analysis, written from the advisory, the CVSS vector and the affected-version data. It adds context the advisory leaves out, and never invents facts that are not in the source.
dbcve analysisData overflows a buffer allocated on the heap, corrupting neighbouring heap structures and allocator metadata that a patient attacker can groom into control of execution. It is subtler than a stack overflow but just as dangerous. The fix is validating lengths before every write and using safe allocators and bounded operations.
General guidance for the heap-based buffer overflow class — the official description and references above are authoritative for this specific CVE. Want a bespoke review and a reviewed fix? Ask our team →
Affected products & versions What the vendor confirmedThe version ranges the vendor confirmed as vulnerable. If your version sits inside a range here, treat yourself as exposed until you have upgraded.
NVD · CPE data< 10.0.28000.2704CVSS breakdown How the score is builtThe industry scoring standard. It rates how the flaw is reached, what it takes to exploit, and what an attacker gains — the score is derived from those, not the other way round.
From the vector- Attack vector
- Local
- Complexity
- Low
- Privileges
- Low
- User interaction
- None
- Scope
- Unchanged
- Confidentiality
- High
- Integrity
- High
- Availability
- High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Remediation Closing itWhat it takes to close this. Where a vendor fix exists we point at it; where none exists we say so plainly, and can build one. Effort estimates are scoped from the advisory, not from your codebase.
From vendor data10.0.28000.2704
Scan for this in your stack
Free · runs locallyCheck whether your project pulls in CVE-2026-62772 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.
References Go to the primary sourcePrimary sources — vendor advisories, patches and trackers. Where our summary and a reference disagree, the reference wins.
Primary sourcesThe CVSS score of 7.8 for CVE-2026-62772 materially understates the risk because this vulnerability exists in a container isolation filter driver — a component whose failure mode is not merely local privilege escalation but potential container-host boundary collapse. Filter drivers operating in kernel mode sit at the intersection of untrusted container I/O requests and host filesystem operations. A heap overflow in unionfs.sys doesn't just let a container workload become root on its own VM — it could let that workload escape to the host kernel. The CVSS metrics treat this as a standard local EOP, but container isolation drivers are trusted components whose compromise invalidates the fundamental security boundary they exist to enforce. The 'authorized attacker' language in the description is particularly dangerous for container threat models: authorization is enforced BY the isolation boundary we're worried about bypassing. Treating a container workload as 'authorized' to exploit a filter driver flaw is like treating a process as 'authorized' to escape its sandbox — it assumes the very integrity the vulnerability questions. The 26H1 designation is also concerning: this is a forward-looking release branch, meaning the vulnerability code is entering deployment pipelines now rather than being retired from legacy systems. This code will likely be backported into mainstream Windows releases and LTS container host images without re-review, compounding its persistence. Microsoft has published seventeen advisories between 2015 and 2024 addressing heap management errors in filesystem filter drivers — the same pattern, the same root cause — yet the identical flaw appears in container isolation drivers. This suggests institutional failure rather than impossible engineering requirements. HVCI addresses exploitation technique, not root cause: a heap overflow that can't be reliably exploited via known HVCI techniques still exists in the codebase. Assess whether Azure Container Instances, Windows Server Containers, and WSL2 are affected. The question isn't how severe this CVE is on its own terms — it's what percentage of your container fleet depends on this filter driver, because CVSS treats vulnerabilities as independent events while container infrastructure treats filter drivers as load-bearing walls.
Practitioner notes
ContributedPeer-ranked notes from engineers who’ve handled CVE-2026-62772 in production — separate from our analysis above.
The advisory tells you what broke. It rarely tells you what actually worked. If you’ve dealt with this one, that detail is what the next engineer is searching for.
- The version that genuinely resolved it — not the one the vendor claimed
- A config change or rule that shut the vector down
- A gotcha in the upgrade path that cost you an afternoon
A place for practitioners to share what actually worked: a mitigation you’ve tested, a configuration change, a version- or environment-specific caveat, or a link to a verified patch. The most useful notes rise to the top as peers upvote them, so the signal stays high.
- Verified mitigations, workarounds, and config changes
- Version or environment caveats, and links to real fixes
- No weaponised exploit code, or anything meant to cause harm
- No spam, self-promotion, credentials, or personal data