Issue with containerd CRI Plugin - CVE-2026-50195, CVE-2026-53488, CVE-2026-53492, CVE-2026-53489, CVE-2026-47262
Multiple critical and high severity vulnerabilities have been identified in the containerd CRI plugin used by Kubernetes runtimes, including GKE and AWS managed services. These vulnerabilities allow attackers with permissions to create Pods to bypass Kubernetes security boundaries, leading to potential host compromise, cache poisoning, arbitrary command execution, unauthorized host file read, and denial of service. The issues affect containerd versions 1.7, 2.0, 2.1, 2.2, and 2.3. Patches are available for Container-Optimized OS node images in GKE and upstream containerd releases. Ubuntu node image patches are still in progress. Mitigations include restricting Pod creation permissions, avoiding certain image pull policies, and limiting checkpoint/restore features to trusted users and images.
AI Analysis
Technical Summary
Five vulnerabilities were discovered in the containerd CRI plugin affecting versions 1.7 through 2.3. CVE-2026-50195 allows image cache poisoning via unvalidated checkpoint image references, enabling cross-pod code execution. CVE-2026-53488 propagates image config labels without validation, allowing arbitrary host command execution. CVE-2026-53492 improperly trusts CDI annotations during container restoration, permitting injection of host mounts and device nodes. CVE-2026-53489 restores container logs without validating symlink paths, enabling arbitrary host file reads. CVE-2026-47262 causes memory exhaustion leading to denial of service. These vulnerabilities require cluster privileges to create Pods for exploitation. Patches are available in upstream containerd releases and GKE Container-Optimized OS node images from versions 1.30 to 1.36. Ubuntu node image patches are pending. AWS is deploying patched runtimes for managed services. Mitigations include restricting Pod creation permissions, using Always image pull policy for untrusted workloads, and restricting checkpoint/restore usage.
Potential Impact
Attackers with permissions to create Pods can bypass Kubernetes security boundaries, leading to host compromise, arbitrary code execution, cache poisoning, unauthorized host file access, and denial of service on nodes running vulnerable containerd versions. These vulnerabilities affect containerd used in Kubernetes runtimes including GKE and AWS managed container services. The requirement for cluster privileges to create Pods reduces the risk to some extent but does not eliminate it. The vulnerabilities impact containerd versions 1.7, 2.0, 2.1, 2.2, and 2.3. GKE Sandbox (gVisor) does not mitigate these host compromise vectors.
Mitigation Recommendations
Patches are available for containerd upstream and for GKE Container-Optimized OS node images from versions 1.30 to 1.36. Users should upgrade to these patched versions or later. Ubuntu node image patches are still in progress; until available, restrict Pod creation permissions to trusted users only, avoid using IfNotPresent or Never image pull policies for untrusted workloads (use Always instead), and restrict or monitor checkpoint/restore features. For CVE-2026-53492, on nodes without CDI usage, remove or relocate host CDI specifications from default directories to prevent exploitation. AWS is deploying patched runtimes for managed services; self-managed deployments should upgrade containerd promptly.
Issue with containerd CRI Plugin - CVE-2026-50195, CVE-2026-53488, CVE-2026-53492, CVE-2026-53489, CVE-2026-47262
Description
Multiple critical and high severity vulnerabilities have been identified in the containerd CRI plugin used by Kubernetes runtimes, including GKE and AWS managed services. These vulnerabilities allow attackers with permissions to create Pods to bypass Kubernetes security boundaries, leading to potential host compromise, cache poisoning, arbitrary command execution, unauthorized host file read, and denial of service. The issues affect containerd versions 1.7, 2.0, 2.1, 2.2, and 2.3. Patches are available for Container-Optimized OS node images in GKE and upstream containerd releases. Ubuntu node image patches are still in progress. Mitigations include restricting Pod creation permissions, avoiding certain image pull policies, and limiting checkpoint/restore features to trusted users and images.
Affected software
pkg:github/containerd/containerdRun on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
Five vulnerabilities were discovered in the containerd CRI plugin affecting versions 1.7 through 2.3. CVE-2026-50195 allows image cache poisoning via unvalidated checkpoint image references, enabling cross-pod code execution. CVE-2026-53488 propagates image config labels without validation, allowing arbitrary host command execution. CVE-2026-53492 improperly trusts CDI annotations during container restoration, permitting injection of host mounts and device nodes. CVE-2026-53489 restores container logs without validating symlink paths, enabling arbitrary host file reads. CVE-2026-47262 causes memory exhaustion leading to denial of service. These vulnerabilities require cluster privileges to create Pods for exploitation. Patches are available in upstream containerd releases and GKE Container-Optimized OS node images from versions 1.30 to 1.36. Ubuntu node image patches are pending. AWS is deploying patched runtimes for managed services. Mitigations include restricting Pod creation permissions, using Always image pull policy for untrusted workloads, and restricting checkpoint/restore usage.
Potential Impact
Attackers with permissions to create Pods can bypass Kubernetes security boundaries, leading to host compromise, arbitrary code execution, cache poisoning, unauthorized host file access, and denial of service on nodes running vulnerable containerd versions. These vulnerabilities affect containerd used in Kubernetes runtimes including GKE and AWS managed container services. The requirement for cluster privileges to create Pods reduces the risk to some extent but does not eliminate it. The vulnerabilities impact containerd versions 1.7, 2.0, 2.1, 2.2, and 2.3. GKE Sandbox (gVisor) does not mitigate these host compromise vectors.
Mitigation Recommendations
Patches are available for containerd upstream and for GKE Container-Optimized OS node images from versions 1.30 to 1.36. Users should upgrade to these patched versions or later. Ubuntu node image patches are still in progress; until available, restrict Pod creation permissions to trusted users only, avoid using IfNotPresent or Never image pull policies for untrusted workloads (use Always instead), and restrict or monitor checkpoint/restore features. For CVE-2026-53492, on nodes without CDI usage, remove or relocate host CDI specifications from default directories to prevent exploitation. AWS is deploying patched runtimes for managed services; self-managed deployments should upgrade containerd promptly.
Technical Details
- Article Source
- {"url":"https://aws.amazon.com/security/security-bulletins/rss/2026-046-aws/","fetched":true,"fetchedAt":"2026-06-20T00:05:05.746Z","wordCount":424}
- Classification
- {"confidence":0.88,"severitySource":"heuristic","classifier":"rss-v2"}
Threat ID: 6a35d932daaa79a87d7037ae
Added to database: 06/20/2026, 00:05:06 UTC
Last enriched: 08/04/2026, 13:03:07 UTC
Last updated: 08/06/2026, 00:41:13 UTC
Views: 719
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
External Links
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.