Container Security vulnerabilities

Docker Model Runner (DMR) is software used to manage, run, and deploy AI models using Docker. Prior to version 1.1.25, Docker Model Runner contains an SSRF vulnerability in its OCI registry token exchange flow. When pulling a model, Model Runner follows the realm URL from the registry's WWW-Authenticate header without validating the scheme, hostname, or IP range. A malicious OCI registry can set the realm to an internal URL (e.g., http://127.0.0.1:3000/), causing Model Runner running on the host to make arbitrary GET requests to internal services and reflect the full response body back to the caller. Additionally, the token exchange mechanism can relay data from internal services back to the attacker-controlled registry via the Authorization: Bearer header. This issue has been patched in version 1.1.25. For Docker Desktop users, enabling Enhanced Container Isolation (ECI) blocks container access to Model Runner, preventing exploitation. However, if the Docker Model Runner is exposed to localhost over TCP in specific configurations, the vulnerability is still exploitable.

Moby is an open source container framework. Prior to version 29.3.1, a security vulnerability has been detected that allows attackers to bypass authorization plugins (AuthZ). This issue has been patched in version 29.3.1.

Moby is an open source container framework. Prior to version 29.3.1, a security vulnerability has been detected that allows plugins privilege validation to be bypassed during docker plugin install. Due to an error in the daemon's privilege comparison logic, the daemon may incorrectly accept a privilege set that differs from the one approved by the user. Plugins that request exactly one privilege are also affected, because no comparison is performed at all. This issue has been patched in version 29.3.1.

Trivy is a security scanner. On March 19, 2026, a threat actor used compromised credentials to publish a malicious Trivy v0.69.4 release, force-push 76 of 77 version tags in `aquasecurity/trivy-action` to credential-stealing malware, and replace all 7 tags in `aquasecurity/setup-trivy` with malicious commits. This incident is a continuation of the supply chain attack that began in late February 2026. Following the initial disclosure on March 1, credential rotation was performed but was not atomic (not all credentials were revoked simultaneously). The attacker could have use a valid token to exfiltrate newly rotated secrets during the rotation window (which lasted a few days). This could have allowed the attacker to retain access and execute the March 19 attack. Affected components include the `aquasecurity/trivy` Go / Container image version 0.69.4, the `aquasecurity/trivy-action` GitHub Action versions 0.0.1 – 0.34.2 (76/77), and the`aquasecurity/setup-trivy` GitHub Action versions 0.2.0 – 0.2.6, prior to the recreation of 0.2.6 with a safe commit. Known safe versions include versions 0.69.2 and 0.69.3 of the Trivy binary, version 0.35.0 of trivy-action, and version 0.2.6 of setup-trivy. Additionally, take other mitigations to ensure the safety of secrets. If there is any possibility that a compromised version ran in one's environment, all secrets accessible to affected pipelines must be treated as exposed and rotated immediately. Check whether one's organization pulled or executed Trivy v0.69.4 from any source. Remove any affected artifacts immediately. Review all workflows using `aquasecurity/trivy-action` or `aquasecurity/setup-trivy`. Those who referenced a version tag rather than a full commit SHA should check workflow run logs from March 19–20, 2026 for signs of compromise. Look for repositories named `tpcp-docs` in one's GitHub organization. The presence of such a repository may indicate that the fallback exfiltration mechanism was triggered and secrets were successfully stolen. Pin GitHub Actions to full, immutable commit SHA hashes, don't use mutable version tags.

A vulnerability was discovered in the Kubernetes CSI Driver for NFS where the subDir parameter in volume identifiers was insufficiently validated. Attackers with the ability to create PersistentVolumes referencing the NFS CSI driver could craft volume identifiers containing path traversal sequences (../). During volume deletion or cleanup operations, the driver could operate on unintended directories outside the intended managed path within the NFS export. This may lead to deletion or modification of directories on the NFS server.

A security issue was discovered in ingress-nginx where a combination of Ingress annotations can be used to inject configuration into nginx. This can lead to arbitrary code execution in the context of the ingress-nginx controller, and disclosure of Secrets accessible to the controller. (Note that in the default installation, the controller can access all Secrets cluster-wide.)

A container privilege escalation flaw was found in certain Multi-Cloud Object Gateway Core images. This issue stems from the /etc/passwd file being created with group-writable permissions during build time. In certain conditions, an attacker who can execute commands within an affected container, even as a non-root user, can leverage their membership in the root group to modify the /etc/passwd file. This could allow the attacker to add a new user with any arbitrary UID, including UID 0, leading to full root privileges within the container

A security issue was discovered in ingress-nginx where the `nginx.ingress.kubernetes.io/rewrite-target` Ingress annotation can be used to inject configuration into nginx. This can lead to arbitrary code execution in the context of the ingress-nginx controller, and disclosure of Secrets accessible to the controller. (Note that in the default installation, the controller can access all Secrets cluster-wide.)

Docker CLI for Windows searches for plugin binaries in C:\ProgramData\Docker\cli-plugins, a directory that does not exist by default. A low-privileged attacker can create this directory and place malicious CLI plugin binaries (docker-compose.exe, docker-buildx.exe, etc.) that are executed when a victim user opens Docker Desktop or invokes Docker CLI plugin features, and allow privilege-escalation if the docker CLI is executed as a privileged user. This issue affects Docker CLI: through 29.1.5 and Windows binaries acting as a CLI-plugin manager using the github.com/docker/cli/cli-plugins/manager https://pkg.go.dev/github.com/docker/cli@v29.1.5+incompatible/cli-plugins/manager  package, such as Docker Compose. This issue does not impact non-Windows binaries, and projects not using the plugin-manager code.

Docker Model Runner (DMR) is software used to manage, run, and deploy AI models using Docker. Versions prior to 1.0.16 expose a POST `/engines/_configure` endpoint that accepts arbitrary runtime flags without authentication. These flags are passed directly to the underlying inference server (llama.cpp). By injecting the --log-file flag, an attacker with network access to the Model Runner API can write or overwrite arbitrary files accessible to the Model Runner process. When bundled with Docker Desktop (where Model Runner is enabled by default since version 4.46.0), it is reachable from any default container at model-runner.docker.internal without authentication. In this context, the file overwrite can target the Docker Desktop VM disk (`Docker.raw` ), resulting in the destruction of all containers, images, volumes, and build history. However, in specific configurations and with user interaction, it is possible to convert this vulnerability in a container escape. The issue is fixed in Docker Model Runner 1.0.16. Docker Desktop users should update to 4.61.0 or later, which includes the fixed Model Runner. A workaround is available. For Docker Desktop users, enabling Enhanced Container Isolation (ECI) blocks container access to Model Runner, preventing exploitation. However, if the Docker Model Runner is exposed to localhost over TCP in specific configurations, the vulnerability is still exploitable.

An out of bounds read vulnerability in the grpcfuse kernel module present in the Linux VM in Docker Desktop for Windows, Linux and macOS up to version 4.61.0 could allow a local attacker to cause an unspecified impact by writing to /proc/docker entries. The issue has been fixed in Docker Desktop 4.62.0 .

cryptography is a package designed to expose cryptographic primitives and recipes to Python developers. Prior to 46.0.5, the public_key_from_numbers (or EllipticCurvePublicNumbers.public_key()), EllipticCurvePublicNumbers.public_key(), load_der_public_key() and load_pem_public_key() functions do not verify that the point belongs to the expected prime-order subgroup of the curve. This missing validation allows an attacker to provide a public key point P from a small-order subgroup. This can lead to security issues in various situations, such as the most commonly used signature verification (ECDSA) and shared key negotiation (ECDH). When the victim computes the shared secret as S = [victim_private_key]P via ECDH, this leaks information about victim_private_key mod (small_subgroup_order). For curves with cofactor > 1, this reveals the least significant bits of the private key. When these weak public keys are used in ECDSA , it's easy to forge signatures on the small subgroup. Only SECT curves are impacted by this. This vulnerability is fixed in 46.0.5.

A security issue was discovered in ingress-nginx where the `nginx.ingress.kubernetes.io/auth-proxy-set-headers` Ingress annotation can be used to inject configuration into nginx. This can lead to arbitrary code execution in the context of the ingress-nginx controller, and disclosure of Secrets accessible to the controller. (Note that in the default installation, the controller can access all Secrets cluster-wide.)

During session resumption in crypto/tls, if the underlying Config has its ClientCAs or RootCAs fields mutated between the initial handshake and the resumed handshake, the resumed handshake may succeed when it should have failed. This may happen when a user calls Config.Clone and mutates the returned Config, or uses Config.GetConfigForClient. This can cause a client to resume a session with a server that it would not have resumed with during the initial handshake, or cause a server to resume a session with a client that it would not have resumed with during the initial handshake.

Docker Desktop for Windows contains multiple incorrect permission assignment vulnerabilities in the installer's handling of the C:\ProgramData\DockerDesktop directory. The installer creates this directory without proper ownership verification, creating two exploitation scenarios: Scenario 1 (Persistent Attack): If a low-privileged attacker pre-creates C:\ProgramData\DockerDesktop before Docker Desktop installation, the attacker retains ownership of the directory even after the installer applies restrictive ACLs. At any time after installation completes, the attacker can modify the directory ACL (as the owner) and tamper with critical configuration files such as install-settings.json to specify a malicious credentialHelper, causing arbitrary code execution when any user runs Docker Desktop. Scenario 2 (TOCTOU Attack): During installation, there is a time-of-check-time-of-use (TOCTOU) race condition between when the installer creates C:\ProgramData\DockerDesktop and when it sets secure ACLs. A low-privileged attacker actively monitoring for the installation can inject malicious files (such as install-settings.json) with attacker-controlled ACLs during this window, achieving the same code execution outcome.

A security issue was discovered in ingress-nginx where the validating admission controller feature is subject to a denial of service condition. By sending large requests to the validating admission controller, an attacker can cause memory consumption, which may result in the ingress-nginx controller pod being killed or the node running out of memory.

A security issue was discovered in ingress-nginx where the protection afforded by the `auth-url` Ingress annotation may not be effective in the presence of a specific misconfiguration. If the ingress-nginx controller is configured with a default custom-errors configuration that includes HTTP errors 401 or 403, and if the configured default custom-errors backend is defective and fails to respect the X-Code HTTP header, then an Ingress with the `auth-url` annotation may be accessed even when authentication fails. Note that the built-in custom-errors backend works correctly. To trigger this issue requires an administrator to specifically configure ingress-nginx with a broken external component.

A security issue was discovered in ingress-nginx where the `rules.http.paths.path` Ingress field can be used to inject configuration into nginx. This can lead to arbitrary code execution in the context of the ingress-nginx controller, and disclosure of Secrets accessible to the controller. (Note that in the default installation, the controller can access all Secrets cluster-wide.)

A security issue was discovered in ingress-nginx where the `nginx.ingress.kubernetes.io/auth-method` Ingress annotation can be used to inject configuration into nginx. This can lead to arbitrary code execution in the context of the ingress-nginx controller, and disclosure of Secrets accessible to the controller. (Note that in the default installation, the controller can access all Secrets cluster-wide.)

Kata Containers is an open source project focusing on a standard implementation of lightweight Virtual Machines (VMs) that perform like containers. In versions prior to 3.26.0, when a container image is malformed or contains no layers, containerd falls back to bind-mounting an empty snapshotter directory for the container rootfs. When the Kata runtime attempts to mount the container rootfs, the bind mount causes the rootfs to be detected as a block device, leading to the underlying device being hotplugged to the guest. This can cause filesystem-level errors on the host due to double inode allocation, and may lead to the host's block device being mounted as read-only. Version 3.26.0 contains a patch for the issue.

Kyverno is a policy engine designed for cloud native platform engineering teams. Versions prior to 1.16.3 and 1.15.3 have a critical authorization boundary bypass in namespaced Kyverno Policy apiCall. The resolved `urlPath` is executed using the Kyverno admission controller ServiceAccount, with no enforcement that the request is limited to the policy’s namespace. As a result, any authenticated user with permission to create a namespaced Policy can cause Kyverno to perform Kubernetes API requests using Kyverno’s admission controller identity, targeting any API path allowed by that ServiceAccount’s RBAC. This breaks namespace isolation by enabling cross-namespace reads (for example, ConfigMaps and, where permitted, Secrets) and allows cluster-scoped or cross-namespace writes (for example, creating ClusterPolicies) by controlling the urlPath through context variable substitution. Versions 1.16.3 and 1.15.3 contain a patch for the vulnerability.

Issue summary: A type confusion vulnerability exists in the signature verification of signed PKCS#7 data where an ASN1_TYPE union member is accessed without first validating the type, causing an invalid or NULL pointer dereference when processing malformed PKCS#7 data. Impact summary: An application performing signature verification of PKCS#7 data or calling directly the PKCS7_digest_from_attributes() function can be caused to dereference an invalid or NULL pointer when reading, resulting in a Denial of Service. The function PKCS7_digest_from_attributes() accesses the message digest attribute value without validating its type. When the type is not V_ASN1_OCTET_STRING, this results in accessing invalid memory through the ASN1_TYPE union, causing a crash. Exploiting this vulnerability requires an attacker to provide a malformed signed PKCS#7 to an application that verifies it. The impact of the exploit is just a Denial of Service, the PKCS7 API is legacy and applications should be using the CMS API instead. For these reasons the issue was assessed as Low severity. The FIPS modules in 3.5, 3.4, 3.3 and 3.0 are not affected by this issue, as the PKCS#7 parsing implementation is outside the OpenSSL FIPS module boundary. OpenSSL 3.6, 3.5, 3.4, 3.3, 3.0, 1.1.1 and 1.0.2 are vulnerable to this issue.

Issue summary: An invalid or NULL pointer dereference can happen in an application processing a malformed PKCS#12 file. Impact summary: An application processing a malformed PKCS#12 file can be caused to dereference an invalid or NULL pointer on memory read, resulting in a Denial of Service. A type confusion vulnerability exists in PKCS#12 parsing code where an ASN1_TYPE union member is accessed without first validating the type, causing an invalid pointer read. The location is constrained to a 1-byte address space, meaning any attempted pointer manipulation can only target addresses between 0x00 and 0xFF. This range corresponds to the zero page, which is unmapped on most modern operating systems and will reliably result in a crash, leading only to a Denial of Service. Exploiting this issue also requires a user or application to process a maliciously crafted PKCS#12 file. It is uncommon to accept untrusted PKCS#12 files in applications as they are usually used to store private keys which are trusted by definition. For these reasons, the issue was assessed as Low severity. The FIPS modules in 3.5, 3.4, 3.3 and 3.0 are not affected by this issue, as the PKCS12 implementation is outside the OpenSSL FIPS module boundary. OpenSSL 3.6, 3.5, 3.4, 3.3, 3.0 and 1.1.1 are vulnerable to this issue. OpenSSL 1.0.2 is not affected by this issue.

Issue summary: The 'openssl dgst' command-line tool silently truncates input data to 16MB when using one-shot signing algorithms and reports success instead of an error. Impact summary: A user signing or verifying files larger than 16MB with one-shot algorithms (such as Ed25519, Ed448, or ML-DSA) may believe the entire file is authenticated while trailing data beyond 16MB remains unauthenticated. When the 'openssl dgst' command is used with algorithms that only support one-shot signing (Ed25519, Ed448, ML-DSA-44, ML-DSA-65, ML-DSA-87), the input is buffered with a 16MB limit. If the input exceeds this limit, the tool silently truncates to the first 16MB and continues without signaling an error, contrary to what the documentation states. This creates an integrity gap where trailing bytes can be modified without detection if both signing and verification are performed using the same affected codepath. The issue affects only the command-line tool behavior. Verifiers that process the full message using library APIs will reject the signature, so the risk primarily affects workflows that both sign and verify with the affected 'openssl dgst' command. Streaming digest algorithms for 'openssl dgst' and library users are unaffected. The FIPS modules in 3.5 and 3.6 are not affected by this issue, as the command-line tools are outside the OpenSSL FIPS module boundary. OpenSSL 3.5 and 3.6 are vulnerable to this issue. OpenSSL 3.4, 3.3, 3.0, 1.1.1 and 1.0.2 are not affected by this issue.

Issue summary: If an application using the SSL_CIPHER_find() function in a QUIC protocol client or server receives an unknown cipher suite from the peer, a NULL dereference occurs. Impact summary: A NULL pointer dereference leads to abnormal termination of the running process causing Denial of Service. Some applications call SSL_CIPHER_find() from the client_hello_cb callback on the cipher ID received from the peer. If this is done with an SSL object implementing the QUIC protocol, NULL pointer dereference will happen if the examined cipher ID is unknown or unsupported. As it is not very common to call this function in applications using the QUIC protocol and the worst outcome is Denial of Service, the issue was assessed as Low severity. The vulnerable code was introduced in the 3.2 version with the addition of the QUIC protocol support. The FIPS modules in 3.6, 3.5, 3.4 and 3.3 are not affected by this issue, as the QUIC implementation is outside the OpenSSL FIPS module boundary. OpenSSL 3.6, 3.5, 3.4 and 3.3 are vulnerable to this issue. OpenSSL 3.0, 1.1.1 and 1.0.2 are not affected by this issue.

Issue summary: Parsing CMS AuthEnvelopedData or EnvelopedData message with maliciously crafted AEAD parameters can trigger a stack buffer overflow. Impact summary: A stack buffer overflow may lead to a crash, causing Denial of Service, or potentially remote code execution. When parsing CMS (Auth)EnvelopedData structures that use AEAD ciphers such as AES-GCM, the IV (Initialization Vector) encoded in the ASN.1 parameters is copied into a fixed-size stack buffer without verifying that its length fits the destination. An attacker can supply a crafted CMS message with an oversized IV, causing a stack-based out-of-bounds write before any authentication or tag verification occurs. Applications and services that parse untrusted CMS or PKCS#7 content using AEAD ciphers (e.g., S/MIME (Auth)EnvelopedData with AES-GCM) are vulnerable. Because the overflow occurs prior to authentication, no valid key material is required to trigger it. While exploitability to remote code execution depends on platform and toolchain mitigations, the stack-based write primitive represents a severe risk. The FIPS modules in 3.6, 3.5, 3.4, 3.3 and 3.0 are not affected by this issue, as the CMS implementation is outside the OpenSSL FIPS module boundary. OpenSSL 3.6, 3.5, 3.4, 3.3 and 3.0 are vulnerable to this issue. OpenSSL 1.1.1 and 1.0.2 are not affected by this issue.

Envoy Gateway is an open source project for managing Envoy Proxy as a standalone or Kubernetes-based application gateway. Prior to 1.5.7 and 1.6.2, EnvoyExtensionPolicy Lua scripts executed by Envoy proxy can be used to leak the proxy's credentials. These credentials can then be used to communicate with the control plane and gain access to all secrets that are used by Envoy proxy, e.g. TLS private keys and credentials used for downstream and upstream communication. This vulnerability is fixed in 1.5.7 and 1.6.2.

A half-blind Server Side Request Forgery (SSRF) vulnerability exists in kube-controller-manager when using the in-tree Portworx StorageClass. This vulnerability allows authorized users to leak arbitrary information from unprotected endpoints in the control plane’s host network (including link-local or loopback services).

Docker Desktop diagnostics bundles were found to include expired Hub PATs in log output due to error object serialization. This poses a risk of leaking sensitive information in exported diagnostics, especially when access denied errors occurred.

In the Linux kernel, the following vulnerability has been resolved: net: bridge: fix use-after-free due to MST port state bypass syzbot reported[1] a use-after-free when deleting an expired fdb. It is due to a race condition between learning still happening and a port being deleted, after all its fdbs have been flushed. The port's state has been toggled to disabled so no learning should happen at that time, but if we have MST enabled, it will bypass the port's state, that together with VLAN filtering disabled can lead to fdb learning at a time when it shouldn't happen while the port is being deleted. VLAN filtering must be disabled because we flush the port VLANs when it's being deleted which will stop learning. This fix adds a check for the port's vlan group which is initialized to NULL when the port is getting deleted, that avoids the port state bypass. When MST is enabled there would be a minimal new overhead in the fast-path because the port's vlan group pointer is cache-hot. [1] https://syzkaller.appspot.com/bug?extid=dd280197f0f7ab3917be

MCP Gateway allows easy and secure running and deployment of MCP servers. In versions 0.27.0 and earlier, when MCP Gateway runs in sse or streaming transport mode, it is vulnerable to DNS rebinding. An attacker who can get a victim to visit a malicious website or be served a malicious advertisement can perform browser-based exploitation of MCP servers executing behind the gateway, including manipulating tools or other features exposed by those MCP servers. MCP Gateway is not affected when running in the default stdio mode, which does not listen on network ports. Version 0.28.0 fixes this issue.

containerd is an open-source container runtime. Versions 1.7.28 and below, 2.0.0-beta.0 through 2.0.6, 2.1.0-beta.0 through 2.1.4, and 2.2.0-beta.0 through 2.2.0-rc.1 contain a bug in the CRI Attach implementation where a user can exhaust memory on the host due to goroutine leaks. This issue is fixed in versions 1.7.29, 2.0.7, 2.1.5 and 2.2.0. To workaround this vulnerability, users can set up an admission controller to control accesses to pods/attach resources.

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.

runc is a CLI tool for spawning and running containers according to the OCI specification. In versions 1.2.7 and below, 1.3.0-rc.1 through 1.3.1, 1.4.0-rc.1 and 1.4.0-rc.2 files, runc would not perform sufficient verification that the source of the bind-mount (i.e., the container's /dev/null) was actually a real /dev/null inode when using the container's /dev/null to mask. This exposes two methods of attack: an arbitrary mount gadget, leading to host information disclosure, host denial of service, container escape, or a bypassing of maskedPaths. This issue is fixed in versions 1.2.8, 1.3.3 and 1.4.0-rc.3.

containerd is an open-source container runtime. Versions 0.1.0 through 1.7.28, 2.0.0-beta.0 through 2.0.6, 2.1.0-beta.0 through 2.1.4 and 2.2.0-beta.0 through 2.2.0-rc.1 have an overly broad default permission vulnerability. Directory paths `/var/lib/containerd`, `/run/containerd/io.containerd.grpc.v1.cri` and `/run/containerd/io.containerd.sandbox.controller.v1.shim` were all created with incorrect permissions. This issue is fixed in versions 1.7.29, 2.0.7, 2.1.5 and 2.2.0. Workarounds include updating system administrator permissions so the host can manually chmod the directories to not have group or world accessible permissions, or to run containerd in rootless mode.

This issue was addressed with improved checks. This issue is fixed in Safari 26.1, iOS 26.1 and iPadOS 26.1, macOS Tahoe 26.1, tvOS 26.1, visionOS 26.1, watchOS 26.1. Processing maliciously crafted web content may lead to an unexpected process crash.

tar.Reader does not set a maximum size on the number of sparse region data blocks in GNU tar pax 1.0 sparse files. A maliciously-crafted archive containing a large number of sparse regions can cause a Reader to read an unbounded amount of data from the archive into memory. When reading from a compressed source, a small compressed input can result in large allocations.

Docker Compose trusts the path information embedded in remote OCI compose artifacts. When a layer includes the annotations com.docker.compose.extends or com.docker.compose.envfile, Compose joins the attacker‑supplied value from com.docker.compose.file/com.docker.compose.envfile with its local cache directory and writes the file there. This affects any platform or workflow that resolves remote OCI compose artifacts, Docker Desktop, standalone Compose binaries on Linux, CI/CD runners, cloud dev environments is affected. An attacker can escape the cache directory and overwrite arbitrary files on the machine running docker compose, even if the user only runs read‑only commands such as docker compose config or docker compose ps. This issue is fixed in v2.40.2.

Docker Desktop Installer.exe is vulnerable to DLL hijacking due to insecure DLL search order. The installer searches for required DLLs in the user's Downloads folder before checking system directories, allowing local privilege escalation through malicious DLL placement.This issue affects Docker Desktop: through 4.48.0.

A flaw was found in Red Hat Openshift AI Service. A low-privileged attacker with access to an authenticated account, for example as a data scientist using a standard Jupyter notebook, can escalate their privileges to a full cluster administrator. This allows for the complete compromise of the cluster's confidentiality, integrity, and availability. The attacker can steal sensitive data, disrupt all services, and take control of the underlying infrastructure, leading to a total breach of the platform and all applications hosted on it.

In a hardened Docker environment, with Enhanced Container Isolation ( ECI https://docs.docker.com/enterprise/security/hardened-desktop/enhanced-container-isolation/ ) enabled, an administrator can utilize the command restrictions feature https://docs.docker.com/enterprise/security/hardened-desktop/enhanced-container-isolation/config/#command-restrictions  to restrict commands that a container with a Docker socket mount may issue on that socket. Due to a software bug, the configuration to restrict commands was ignored when passed to ECI, allowing any command to be executed on the socket. This grants excessive privileges by permitting unrestricted access to powerful Docker commands. The vulnerability affects only Docker Desktop 4.46.0 users that have ECI enabled and are using the Docker socket command restrictions feature. In addition, since ECI restricts mounting the Docker socket into containers by default, it only affects containers which are explicitly allowed by the administrator to mount the Docker socket.

A vulnerability exists in the Kubernetes C# client where the certificate validation logic accepts properly constructed certificates from any Certificate Authority (CA) without properly verifying the trust chain. This flaw allows a malicious actor to present a forged certificate and potentially intercept or manipulate communication with the Kubernetes API server, leading to possible man-in-the-middle attacks and API impersonation.

Kubernetes secrets-store-sync-controller in versions before 0.0.2 discloses service account tokens in logs.

Envoy is an open source L7 proxy and communication bus designed for large modern service oriented architectures. Versions 1.34.0 through 1.34.4 and 1.35.0 contain a use-after-free (UAF) vulnerability in the DNS cache, causing abnormal process termination. The vulnerability is in Envoy's Dynamic Forward Proxy implementation, occurring when a completion callback for a DNS resolution triggers new DNS resolutions or removes existing pending resolutions. This condition may occur when the following conditions are met: dynamic Forwarding Filter is enabled, the `envoy.reloadable_features.dfp_cluster_resolves_hosts` runtime flag is enabled, and the Host header is modified between the Dynamic Forwarding Filter and Router filters. This issue is resolved in versions 1.34.5 and 1.35.1. To work around this issue, set the envoy.reloadable_features.dfp_cluster_resolves_hosts runtime flag to false.

A vulnerability was identified in Docker Desktop that allows local running Linux containers to access the Docker Engine API via the configured Docker subnet, at 192.168.65.7:2375 by default. This vulnerability occurs with or without Enhanced Container Isolation (ECI) enabled, and with or without the "Expose daemon on tcp://localhost:2375 without TLS" option enabled. This can lead to execution of a wide range of privileged commands to the engine API, including controlling other containers, creating new ones, managing images etc. In some circumstances (e.g. Docker Desktop for Windows with WSL backend) it also allows mounting the host drive with the same privileges as the user running Docker Desktop.

A security issue was discovered in the Kubernetes Image Builder where default credentials are enabled during the Windows image build process when using the Nutanix or VMware OVA providers. These credentials, which allow root access, are disabled at the conclusion of the build. Kubernetes clusters are only affected if their nodes use VM images created via the Image Builder project and the vulnerability was exploited during the build process, which requires an attacker to access the build VM and modify the image while the build is in progress.

NVIDIA Container Toolkit for all platforms contains a vulnerability in some hooks used to initialize the container, where an attacker could execute arbitrary code with elevated permissions. A successful exploit of this vulnerability might lead to escalation of privileges, data tampering, information disclosure, and denial of service.

System environment variables are recorded in Docker Desktop diagnostic logs, when using shell auto-completion. This leads to unintentional disclosure of sensitive information such as api keys, passwords, etc.  A malicious actor with read access to these logs could obtain secrets and further use them to gain unauthorized access to other systems. Starting with version 4.43.0 Docker Desktop no longer logs system environment variables as part of diagnostics log collection.

A vulnerability exists in the NodeRestriction admission controller where nodes can bypass dynamic resource allocation authorization checks. When the DynamicResourceAllocation feature gate is enabled, the controller properly validates resource claim statuses during pod status updates but fails to perform equivalent validation during pod creation. This allows a compromised node to create mirror pods that access unauthorized dynamic resources, potentially leading to privilege escalation.