Red Hat Red Hat OpenShift AI (RHOAI) vulnerabilities

A flaw was found in jwcrypto. The JWK.import_key() function validates the key_ops JWK member for duplicate values using an algorithm with O(n^2) time complexity, and the length of key_ops is not bounded. A remote, unauthenticated attacker can supply a JWK with a large key_ops array to an application that passes attacker-controlled key material to a public key-import API (reachable via ECDH-ES key agreement, OIDC dynamic client registration, DPoP, or ACME account key registration, among others) to consume excessive CPU time, resulting in a denial of service.

A flaw was found in odh-dashboard in Red Hat OpenShift AI. The backend-for-frontend route GET /api/nim-serving/:nimResource reads Kubernetes Secrets using the dashboard service account and returns the full Secret object, including .data, without an authorization check. Any authenticated dashboard user can retrieve the cluster NVIDIA NGC API key Secret (apiKeySecret) and the NIM image pull secret (nimPullSecret). Create and delete of the same NIM credential are admin-gated; the read path is not. This is missing authorization (CWE-862) and insufficiently protected credentials (CWE-522). It is distinct from CVE-2026-5483 (service-account token leak in the Kubernetes client response wrapper on the same route) and CVE-2026-16456 (odh-model-controller cross-namespace confused deputy).

A flaw was found in FFmpeg. The tdsc_load_cursor() function writes beyond the bounds of a heap-allocated buffer when processing crafted TDSC cursor data. A remote attacker could exploit this by supplying a specially crafted video file, potentially leading to a denial of service or arbitrary code execution.

A flaw was found in jwcrypto. A remote attacker can send a specially crafted JSON Web Encryption (JWE) token containing numerous period delimiters. This malformed token can force the JWE.deserialize() function to allocate excessive memory, leading to a MemoryError. This issue results in a denial of service (DoS) for services that process untrusted JWE values.

A flaw was found in the gorch service template, which is part of the trustyai-service-operator. Even when authentication is enabled, the gorch service exposes unproxied orchestrator and detector metrics ports. This allows any pod on the cluster network to directly access these ports, bypassing the kube-rbac-proxy and its authentication mechanisms. This could lead to unauthorized access to the orchestrator and detector metrics.

A flaw was found in the TrustyAI Service Operator. When deploying services like gorch or NemoGuardrails, if a specific security setting is not enabled, these services can expose their communication channels without requiring users to prove their identity. This allows any other program within the cluster to access the AI guardrails and orchestrator without proper authorization. An attacker could exploit this to gain unauthorized access to sensitive information and potentially make limited changes to the AI models.

A use-after-free vulnerability was found in FFmpeg's RASC video decoder. The decode_move() function initializes a read pointer into a decompressed buffer, but a subsequent reallocation of that same buffer during move-table processing leaves the pointer dangling. An attacker could exploit this by providing a specially crafted AVI file containing a malicious RASC video stream. When a user opens or plays the file, the decoder reads from freed heap memory, which could lead to a denial of service (crash).

A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from improper handling of image metadata, specifically EXIF orientation and PNG transparency (tRNS) data, during image processing. When images are converted to RGB, transparency information may be implicitly discarded or remapped, leading to unexpected rendering of transparent pixels and distortion of input content. This can result in the model misinterpreting image content, potentially affecting the integrity of processed data.

A flaw was found in FFmpeg. A remote attacker could exploit this vulnerability by providing a specially crafted MPEG-PS/VOB media file containing a malicious DVD subtitle stream. This vulnerability is caused by a signed integer overflow in the DVD subtitle parser's fragment reassembly bounds checks, leading to a heap out-of-bounds write. Successful exploitation can result in a denial of service (DoS) due to an application crash, and potentially lead to arbitrary code execution.

A flaw was found in uv. This vulnerability allows an attacker to execute malicious code during package resolution or installation via specially crafted ZIP (Zipped Information Package) archives that exploit parsing differentials, requiring user interaction to install an attacker-controlled package.

A flaw was found in github.com/go-viper/mapstructure/v2, in the field processing component using mapstructure.WeakDecode. This vulnerability allows information disclosure through detailed error messages that may leak sensitive input values via malformed user-supplied data processed in security-critical contexts.

A flaw was found in Red Hat Openshift AI Service. The TrustyAI component is granting all service accounts and users on a cluster permissions to get, list, watch any pod in any namespace on the cluster. TrustyAI is creating a role `trustyai-service-operator-lmeval-user-role` and a CRB `trustyai-service-operator-default-lmeval-user-rolebinding` which is being applied to `system:authenticated` making it so that every single user or service account can get a list of pods running in any namespace on the cluster Additionally users can access all `persistentvolumeclaims` and `lmevaljobs`

A command injection vulnerability was discovered in the TrustyAI Explainability toolkit. Arbitrary commands placed in certain fields of a LMEValJob custom resource (CR) may be executed in the LMEvalJob pod's terminal. This issue can be exploited via a maliciously crafted LMEvalJob by a user with permissions to deploy a CR.

A flaw was found in npm-serialize-javascript. The vulnerability occurs because the serialize-javascript module does not properly sanitize certain inputs, such as regex or other JavaScript object types, allowing an attacker to inject malicious code. This code could be executed when deserialized by a web browser, causing Cross-site scripting (XSS) attacks. This issue is critical in environments where serialized data is sent to web clients, potentially compromising the security of the website or web application using this package.