A critical vulnerability in containerd allows malicious image labels to trigger arbitrary command execution on the host, threatening the security of the entire containerized environment.

The CRI plugin in containerd propagates image labels from a Dockerfile directly to the container runtime without adequate validation. If a secondary plugin or component consumes these labels to perform operations, an attacker can craft malicious labels to break out of the container boundary and execute commands on the host.

With a CVSS score of 9.4, this flaw poses an extreme risk to multi-tenant or shared-hosting container environments. Exploitation allows for container escape, resulting in full host compromise, cross-container data access, and the potential to disrupt all workloads running on the affected node.

Immediate Action: Update containerd to the patched versions (1.7.33, 2.3.2, 2.2.5, 2.1.9, or 2.0.10) immediately.

Proactive Monitoring: Review container image registry logs for unusual LABEL instructions and monitor host system logs for unexpected process execution originating from container runtimes.

Compensating Controls: Implement strict image signing and validation policies (e.g., Notary/Cosign) to ensure that only trusted images are deployed and to prevent the injection of malicious labels.

Public Exploit Available: No

Container escape vulnerabilities are high-value targets for attackers. It is imperative that all container hosts be updated to the specified versions to prevent host-level compromise. Organizations should also audit their container build pipelines to ensure that only authorized images are permitted for deployment.