CVE-2026-23968 is a medium severity vulnerability in the Copier library and CLI tool used for rendering project templates. Versions prior to 9.11.2 allow safe templates to include arbitrary files or directories outside the intended template location by exploiting symbolic link (symlink) following behavior combined with the default setting `_preserve_symlinks: false`. This can lead to unintended file disclosure or manipulation during template generation. The vulnerability does not require privileges or authentication but does require user interaction to trigger template rendering. The issue is patched in version 9.11.2. European organizations using Copier versions before 9.
CVE-2025-33225 is a high-severity vulnerability in the NVIDIA Resiliency Extension for Linux, affecting versions prior to 0.5.0. It involves a symbolic link (symlink) following weakness in log aggregation, where attackers can predict log file names. Exploiting this flaw can lead to privilege escalation, arbitrary code execution, denial of service, information disclosure, and data tampering without requiring user interaction or prior authentication. The vulnerability has a CVSS score of 8.4, indicating significant risk. No known exploits are currently reported in the wild. European organizations using NVIDIA Resiliency Extension in Linux environments should prioritize patching once available and implement strict file system permissions and monitoring to mitigate risks. Countries with strong tech sectors and high adoption of NVIDIA products, such as Germany, France, and the UK, are most likely to be affected.
CVE-2025-64750 is a medium severity vulnerability in Sylabs Singularity container platforms (SingularityCE and SingularityPRO) prior to versions 4.3.5 and 4.1.11. It involves a symbolic link (symlink) following issue where an attacker can bypass Linux Security Module (LSM) restrictions by redirecting the mount of /proc to a shared mount controlled by the attacker. Exploitation requires the victim to run a malicious container image that mounts /proc to this attacker-controlled location. The attacker must also have control over the shared mount content, either via another malicious container or host permissions. This vulnerability can lead to partial compromise of confidentiality, integrity, and availability of the container environment. No known exploits are reported in the wild.
Open OnDemand is an open-source HPC portal. Prior to versions 4.0.8 and 3.1.16, users can craft a "Time of Check to Time of Use" (TOCTOU) attack when downloading zip files to access files outside of the OOD_ALLOWLIST. This vulnerability impacts sites that use the file browser allowlists in all current versions of OOD. However, files accessed are still protected by the UNIX permissions. Open OnDemand versions 4.0.8 and 3.1.16 have been patched for this vulnerability.
runc is a CLI tool for spawning and running containers according to the OCI specification. In versions 1.2.7, 1.3.2 and 1.4.0-rc.2, an attacker can trick runc into misdirecting writes to /proc to other procfs files through the use of a racing container with shared mounts (we have also verified this attack is possible to exploit using a standard Dockerfile with docker buildx build as that also permits triggering parallel execution of containers with custom shared mounts configured). This redirect could be through symbolic links in a tmpfs or theoretically other methods such as regular bind-mounts. While similar, the mitigation applied for the related CVE, CVE-2019-19921, was fairly limited and effectively only caused runc to verify that when LSM labels are written they are actually procfs files. This issue is fixed in versions 1.2.8, 1.3.3, and 1.4.0-rc.3.
runc is a CLI tool for spawning and running containers according to the OCI specification. Versions 1.0.0-rc3 through 1.2.7, 1.3.0-rc.1 through 1.3.2, and 1.4.0-rc.1 through 1.4.0-rc.2, due to insufficient checks when bind-mounting `/dev/pts/$n` to `/dev/console` inside the container, an attacker can trick runc into bind-mounting paths which would normally be made read-only or be masked onto a path that the attacker can write to. This attack is very similar in concept and application to CVE-2025-31133, except that it attacks a similar vulnerability in a different target (namely, the bind-mount of `/dev/pts/$n` to `/dev/console` as configured for all containers that allocate a console). This happens after `pivot_root(2)`, so this cannot be used to write to host files directly -- however, as with CVE-2025-31133, this can load to denial of service of the host or a container breakout by providing the attacker with a writable copy of `/proc/sysrq-trigger` or `/proc/sys/kernel/core_pattern` (respectively). This issue is fixed in versions 1.2.8, 1.3.3 and 1.4.0-rc.3.
CVE-2025-31133 is a vulnerability in the runc container runtime tool where insufficient verification of the source of bind-mounts for /dev/null allows attackers to exploit symbolic link following. This flaw enables arbitrary mount gadget attacks that can lead to host information disclosure, denial of service, container escape, or bypassing of maskedPaths protections. The issue affects specific versions of runc prior to 1.2.8, between 1.3.0-rc.1 and before 1.3.3, and certain 1.4.0 release candidates. It has a high severity rating and a CVSS 4.0 score of 7.3. Fixes are available in versions 1.2.8, 1.3.3, and 1.4.0-rc.3.
Youki is a container runtime written in Rust. In versions 0.5.6 and below, youki’s apparmor handling performs insufficiently strict write-target validation, and when combined with path substitution during pathname resolution, can allow writes to unintended procfs locations. While resolving a path component-by-component, a shared-mount race can substitute intermediate components and redirect the final target. This issue is fixed in version 0.5.7.
Youki is a container runtime written in Rust. In versions 0.5.6 and below, the initial validation of the source /dev/null is insufficient, allowing container escape when youki utilizes bind mounting the container's /dev/null as a file mask. This issue is fixed in version 0.5.7.
SupportAssist for Home PCs versions 4.8.2 and prior and SupportAssist for Business PCs versions 4.5.3 and prior, contain an UNIX Symbolic Link (Symlink) following vulnerability. A low privileged attacker with local access to the system could potentially exploit this vulnerability to delete arbitrary files only in that affected system.
Claude Code is an agentic coding tool. Versions below 1.0.120 failed to account for symlinks when checking permission deny rules. If a user explicitly denied Claude Code access to a file and Claude Code had access to a symlink pointing to that file, it was possible for Claude Code to access the file. Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to the latest version. This issue is fixed in version 1.0.120.
A UNIX Symbolic Link (Symlink) Following vulnerability in logrotate config in the exim package allowed privilege escalation from mail user/group to root.This issue affects Tumbleweed: from ? before 4.98.2-lp156.248.1.
tar-fs provides filesystem bindings for tar-stream. Versions prior to 3.1.1, 2.1.3, and 1.16.5 are vulnerable to symlink validation bypass if the destination directory is predictable with a specific tarball. This issue has been patched in version 3.1.1, 2.1.4, and 1.16.6. A workaround involves using the ignore option on non files/directories.
astral-tokio-tar is a tar archive reading/writing library for async Rust. In versions 0.5.3 and earlier of astral-tokio-tar, tar archives may extract outside of their intended destination directory when using the Entry::unpack_in_raw API. Additionally, the Entry::allow_external_symlinks control (which defaults to true) could be bypassed via a pair of symlinks that individually point within the destination but combine to point outside of it. These behaviors could be used individually or combined to bypass the intended security control of limiting extraction to the given directory. This in turn would allow an attacker with a malicious tar archive to perform an arbitrary file write and potentially pivot into code execution. This issue has been patched in version 0.5.4. There is no workaround other than upgrading.
The txtai framework allows the loading of compressed tar files as embedding indices. While the validate function is intended to prevent path traversal vulnerabilities by ensuring safe filenames, it does not account for symbolic links within the tar file. An attacker is able to write a file anywhere in the filesystem when txtai is used to load untrusted embedding indices
A UNIX Symbolic Link (Symlink) Following vulnerability in the packaging of openSUSE Tumbleweed traefik2 allows the traefik user to escalate to root. This issue affects Tumbleweed: from ? before 2.11.29.
Airlink's Daemon interfaces with Docker and the Panel to provide secure access for controlling instances via the Panel. In version 1.0.0, an attacker with access to the affected container can create symbolic links inside the mounted directory (/app/data). Because the container bind-mounts an arbitrary host path, these symlinks can point to sensitive locations on the host filesystem. When the application or other processes follow these symlinks, the attacker can gain unauthorized read access to host files outside the container. This issue has been patched in version 1.0.1.
Youki is a container runtime written in Rust. Prior to version 0.5.5, if /proc and /sys in the rootfs are symbolic links, they can potentially be exploited to gain access to the host root filesystem. This issue has been patched in version 0.5.5.
Using Codex CLI in workspace-write mode inside a malicious context (repo, directory, etc) could lead to arbitrary file overwrite and potentially remote code execution due to symlinks being followed outside the allowed current working directory.
Improper handling of symbolic links in Ivanti Connect Secure before version 22.7R2.8 or 22.8R2, Ivanti Policy Secure before 22.7R1.5, Ivanti ZTA Gateway before 22.8R2.3-723 and Ivanti Neurons for Secure Access before 22.8R1.4 (Fix deployed on 02-Aug-2025) allows a local authenticated attacker to read arbitrary files on disk.
