CVE-2026-57231: Podman Information Disclosure Vulnerability

CVE-2026-57231 Overview

Podman is an open-source tool for managing Open Container Initiative (OCI) containers and pods. CVE-2026-57231 is an information disclosure vulnerability affecting Podman versions from 1.8.1 up to 5.8.4. A malicious container image can declare an environment variable with only a key and no value, which tricks Podman into forwarding the corresponding host environment variable into the container. When the key is an asterisk (*), Podman forwards every host environment variable present in the launching session. This enables a malicious image to exfiltrate secrets, tokens, and credentials stored in the host shell environment. The issue is fixed in Podman 5.8.4 and 6.0.0.

Critical Impact
A malicious container image can silently exfiltrate all host environment variables from the session that launched the container, exposing secrets, API tokens, and credentials.

Affected Products

Podman versions 1.8.1 through 5.8.3
Podman 5.x releases prior to 5.8.4
Podman 6.x pre-release builds prior to 6.0.0

Discovery Timeline

2026-06-26 - CVE-2026-57231 published to NVD
2026-06-27 - Last updated in NVD database

Technical Details for CVE-2026-57231

Vulnerability Analysis

CVE-2026-57231 is an information disclosure flaw classified under [CWE-200] Exposure of Sensitive Information to an Unauthorized Actor. The defect resides in how Podman processes the ENV directives declared inside a container image. When an image declares an environment variable using a key without a value, Podman treats the entry as a request to inherit that variable from the host process environment. This behavior mirrors shell-style variable passthrough but happens implicitly during container startup.

The severity is amplified when the declared key is a wildcard *. In that case, Podman iterates over the entire host environment and injects every variable into the container. An attacker who publishes a crafted image to a public registry can therefore harvest secrets from any user who runs the image locally.

Root Cause

The root cause lies in Podman's environment variable merging logic. The parser interprets an empty value as an instruction to look up the key in the host environment rather than treating it as an explicit empty string. No filtering exists for wildcard keys, so * expands to the full host environment. Sessions frequently contain sensitive variables such as AWS_SECRET_ACCESS_KEY, GITHUB_TOKEN, KUBECONFIG paths, and CI/CD credentials.

Attack Vector

An attacker publishes a container image containing an ENV directive with a bare key or *. A victim pulls and runs the image with podman run. Podman injects host environment variables into the container process. The container's ENTRYPOINT or CMD reads those variables and transmits them to an attacker-controlled endpoint. Exploitation requires no privileges, no user interaction beyond running the image, and can be triggered remotely by any user who trusts the malicious image source. Refer to the GitHub Security Advisory GHSA-4hq8-gpf5-8p68 for the vendor writeup.

Detection Methods for CVE-2026-57231

Indicators of Compromise

Container images that declare ENV directives with keys but no values, particularly wildcard * entries in the image manifest or Dockerfile.
Outbound network connections from short-lived container processes to unknown domains shortly after image pull.
Podman audit logs showing container launches followed by unexpected DNS queries or HTTP POST requests carrying base64 or JSON payloads.

Detection Strategies

Inspect container images before execution using podman image inspect <image> and review the Config.Env field for empty values or * entries.
Monitor podman run invocations across build agents and developer workstations for images pulled from untrusted registries.
Correlate container process telemetry with parent shell environments to identify unexpected variable propagation.

Monitoring Recommendations

Enable Podman event logging with podman events and forward events to a centralized log platform for review.
Alert on containers that spawn network connections within seconds of starting when running images from public registries.
Baseline expected environment variables per workload and flag deviations that expose secrets to container runtimes.

How to Mitigate CVE-2026-57231

Immediate Actions Required

Upgrade Podman to version 5.8.4 or 6.0.0 on all hosts, including developer laptops, CI runners, and production nodes.
Rotate any secrets that may have been present in shell environments where untrusted images were executed.
Audit shell profiles and CI job definitions to remove long-lived credentials from process environments.

Patch Information

The fix is available in Podman 5.8.4 and 6.0.0. The remediation commit is published in the upstream repository as GitHub commit 6c431b7. The patch corrects the environment merging logic so that keys without values and wildcard * entries are no longer expanded from the host environment.

Workarounds

Only run container images from trusted, signed sources and enforce image signature verification with podman image trust.
Launch Podman from a minimal shell session that strips sensitive variables using env -i podman run ....
Use --env-file with an explicit allowlist instead of relying on implicit host environment inheritance.

bash

# Configuration example
# Run Podman with a clean environment and an explicit allowlist
env -i PATH=/usr/bin:/bin HOME=$HOME \
  podman run --rm \
    --env-file /etc/podman/allowed.env \
    <image>:<tag>

# Verify the installed Podman version includes the fix
podman --version   # Expect 5.8.4 or 6.0.0 or later