CVE-2026-94603: A Podman Image Annotation Switches Off the Sandbox You Asked For

You do not fully trust an image, so you run it with --cap-drop=ALL and move on. On rootful Podman from 4.4.0 until this week's fixes, that image could overrule you with one annotation, and Podman would silently ignore what you asked for.
| Item | Detail |
|---|---|
| Identifier | CVE-2026-94603, GHSA-2cvf-wqm6-wr9g |
| Affected | Podman from 4.4.0 before the fixed releases, rootful containers only |
| Fixed in | 5.8.8 and 6.1.3 |
| Severity | 10.0 Critical, CVSS 3.1, as shown on the Podman project's GitHub advisory |
| Exploited as of 2026-10-01 | None reported in the advisory or the release notes |
| Disclosed | 2026-09-29 |
An image and a checkpoint are not the same thing
Podman 4.4.0 added a feature that let checkpoints from podman container checkpoint travel through an OCI registry. The original commit, written in November 2022, states the behaviour plainly: "When podman run is invoked with an image that contains a checkpoint, it would restore the container from that checkpoint."
That made one command mean two things. Running an image means you decide the sandbox. Restoring a checkpoint means the checkpoint brings its own container configuration. The thing that chose between those two modes was the image itself, through one line in the code that the fix removed:
_, isCheckpointImage := imgData.Annotations[define.CheckpointAnnotationRuntimeName]
That constant is the key io.podman.annotations.checkpoint.runtime.name. The advisory is blunt about the test: "If this annotation is set, the image is treated as a checkpoint." Presence of the key was enough.
When the key was present, podman run called ic.ContainerRestore with options carrying only the name and pod from the run spec, then returned a report straight away. That return sits before generate.MakeContainer in the same function, so your flags never became a container spec. The early return is my reading of the diff. The advisory states the outcome directly: Podman will "ignore --cap-drop=all and use the set of capabilities specified by the checkpoint image instead, which could include all capabilities."
The project's sources differ on how far this goes. The advisory says "almost all sandboxing". The 5.8.8 release notes say it "could disable all sandboxing, including sandboxing specified by the user". Plan for the second wording.
The maintainers name the root issue in one clause: this "allows inherently insecure behavior as the image can now control how it is executed." This is a design flaw, not a memory bug, so there was no parser to harden. They removed the feature, deleting 64 lines across the local and remote code paths in the revert, then added a server-side check, so podman run on an annotated image now fails with an error telling you to use podman container restore. I think that was the right call. Any stricter check on the annotation would still be the image deciding how it runs.
The advisory credits ADA Logics Ltd, using Anthropic's Claude models, with the discovery.
Read the 10.0 against your own setup
The advisory's vector is CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H. It claims no user interaction, yet someone or something has to run the image. That sits awkwardly, and it is why a score alone does not set your priority.
My position is to rank by one question: can an image someone else built reach a rootful podman run on your systems? CI runners and automation that pull third-party images go first. A host running only images you build yourself can follow in normal patching. Neither the advisory nor the release notes report exploitation, but the disclosure is only two days old.
What to check and do now
- Find the exposed hosts. You are looking for rootful Podman at 4.4.0 or later. Rootless containers are not affected, because they "do not support checkpoint/restore at all currently".
- Upgrade to 5.8.8 on the 5.x line or 6.1.3 on the 6.x line, or later. The advisory lists
github.com/containers/podmanpatched at v5.8.8 andgo.podman.io/podman/v6patched at v6.1.3, so check Go projects that import those modules as well. Upstream names no other fixed release. If you run a distribution package, ask your distribution whether it carries the revert instead of judging by the version number. The remote client held its own copy of the check, so upgradepodman-remoteclients along with hosts. - Until then, apply the advisory's workaround. Scan pulled images for
io.podman.annotations.checkpoint.runtime.nameand reject any image that has it. The advisory says Podman "does not presently have the ability to perform this scanning and rejection automatically", so this check has to live in your own pipeline, at a registry mirror or a CI gate. The advisory gives no exact command. The vulnerable code read the annotation from the image's inspect data, sopodman image inspectoutput is where to look on a host. Do not run untrusted images under rootfulpodman rununtil this is in place. - Look backwards. Search registry mirrors, CI caches and hosts for images carrying the key. Then compare the capabilities of running containers against what you requested. That second check is my inference from the described behaviour, not a detection any source documents.
- If you used this feature, expect breakage. The release notes list it as a breaking change: "Removed support for checkpoint images in podman run due to serious security concerns". On the fixed releases,
podman runon such an image now errors rather than starting a container, and the message says the image "must be started usingpodman container restoreinstead". Move to that explicit command. The revert commit says checkpoints "must be explicitly requested by the user" and recommends monitoring a checkpoint's whole lifecycle and ensuring its integrity at each stage.
Image metadata is attacker input
A review of an image's packages has nothing to report here, because the package list can be spotless while the annotation sits beside it. A source scan is looking at a different object altogether. You have to scan the artifact you ship, and that includes its metadata, not only its contents. The advisory's only workaround is an artifact check, which tells you where the control has to sit.
Treat every field in an image as written by whoever built it, and for each tool that reads a field, ask whether that field changes what the tool does. Instructions about how to run a container should come from the operator. Image data should never be able to widen them. If you build tooling that reads annotations, do not let one loosen a security setting. Make the dangerous mode something the user explicitly asks for, which is where Podman has now landed.
Pick one host running rootful Podman today. List the images it has pulled, read the annotations on each, and write down which keys you cannot explain. Start your reject rule with the key above, and grow it from that list.
References
- Podman project, GHSA-2cvf-wqm6-wr9g: The
podman runcommand can be instructed to disable almost all sandboxing by image annotation, 2026-09-29 - Podman project, Release v5.8.8, 2026-09-29
- Podman project, Release v6.1.3, 2026-09-29
- Podman project, Revert of "Enable podman run for checkpoint images", 2026-09-28
- Podman project, Error when a checkpoint image is run, 2026-09-28
- Podman project, Enable podman run for checkpoint images (original feature commit), 2022-11-20
