CVE-2025-23267 — nvidia-container-toolkit
HIGHCVE-2025-23267 is a high-severity (CVSS 8.5) CWE-59 vulnerability in github.com/NVIDIA/nvidia-container-toolkit. A fix is available for github.com/NVIDIA/nvidia-container-toolkit — see the affected versions and patch details below.
NVIDIA Container Toolkit for all platforms contains a vulnerability in the update-ldcache hook
Exploitation Status
No confirmed exploitation observed yet
- CISA’s own triage has not observed active exploitation or public proof-of-concept code for this CVE as of its last assessment.
Exploitation and automatability from CISA’s SSVC triage for CVE-2025-23267.
EPSS Exploitation Probability
Probability of exploitation in the next 30 days, from FIRST.org EPSS.
How urgent is this, really
Where this sits among everything scored
Real-World Exposure
github.com/NVIDIA/nvidia-container-toolkit🐹github.com/NVIDIA/k8s-device-plugin🐹github.com/NVIDIA/gpu-operator🐹github.com/NVIDIA/mig-partedReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects Go packages — download data is not available via public APIs for these ecosystems.
Description
NVIDIA Container Toolkit for all platforms contains a vulnerability in the update-ldcache hook, where an attacker could cause a link following by using a specially crafted container image. A successful exploit of this vulnerability might lead to data tampering and denial of service.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐹Go | github.com/NVIDIA/nvidia-container-toolkit | all versions | 1.17.8go get github.com/NVIDIA/nvidia-container-toolkit@v1.17.8 |
| 🐹Go | github.com/NVIDIA/k8s-device-plugin | all versions | 0.17.3go get github.com/NVIDIA/k8s-device-plugin@v0.17.3 |
| 🐹Go | github.com/NVIDIA/gpu-operator | all versions | 25.3.2go get github.com/NVIDIA/gpu-operator@v25.3.2 |
| 🐹Go | github.com/NVIDIA/mig-parted | all versions | 0.12.2go get github.com/NVIDIA/mig-parted@v0.12.2 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for github.com/NVIDIA/nvidia-container-toolkit, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update github.com/NVIDIA/nvidia-container-toolkit to 1.17.8 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2025-23267 is resolved across your whole dependency graph.
Workarounds
Cap what an attacker can consume: apply request size, rate and timeout limits in front of the affected component, and run it with memory and CPU limits so exhaustion degrades one worker rather than the whole service.
Fixing This On Your OS
If you run this on a Linux distribution, patch through your package manager against the distro's own security advisory below — it tracks the exact backported fix for your release, which can ship on a different timeline (and sometimes a different severity) than the upstream project.
RHEL AI is not affected because it uses CDI mode with NVIDIA Container Toolkit versions later than 1.17.5, which are not vulnerable, and does not rely on the enable-cuda-compat hook, thereby eliminating the attack vector entirely. This vulnerability is rated as Important instead of Critical because, while it involves…
This vulnerability can be mitigated by opting out using the enable-cuda-compat hook. When using the NVIDIA Container Runtime in legacy mode, you can opt out of using the enable-cuda-compat hook by editing the /etc/nvidia-container-toolkit/config.toml file to set the features.disable-cuda-compat-lib-hook feature flag to true: [features] disable-cuda-compat-lib-hook = trueSource: Red Hat security advisory for CVE-2025-23267 (CC BY 4.0)
Frequently Asked Questions
Is CVE-2025-23267 in your dependencies?
Find it across Go, including transitive dependencies.