Data Management and Analytics

In MariaDB Server versions 10.6.1 to before 10.6.27, 10.11.1 to before 10.11.18, 11.4.1 to before 11.4.12, 11.8.1 to before 11.8.8, and 12.3.1 a high severity vulnerability CVE-2026-48165 was detected. This vulnerability allows a high-privileged MariaDB user to execute arbitrary shell commands with the privileges of the mariadbd process on the galera joiner node. This occurs due to improper handling of the wsrep_sst_receive_address or wsrep_sst_donor global system variables. To address this issue, users should upgrade MariaDB Server to versions 10.6.27, 10.11.18, 11.4.12, 11.8.8, or 12.3.2. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-48165.

In LiteLLM versions 1.74.2 to before 1.83.7 a medium severity vulnerability CVE-2026-42271 was detected. This vulnerability allows any authenticated user, including those with low-privileged internal-user keys, to execute arbitrary commands on the proxy host. This occurs because the MCP server preview endpoints (POST /mcp-rest/test/connection and POST /mcp-rest/test/tools/list) improperly accept and execute a full server configuration from the request body. When a stdio configuration containing command, args, and env fields is supplied, the endpoints spawn the specified command as a subprocess with the privileges of the proxy process, completely bypassing role-based access checks. To address this issue, users should upgrade LiteLLM to version 1.83.7. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-42271.

In LiteLLM versions prior to 1.83.10 a high severity vulnerability CVE-2026-47102 was detected. This vulnerability allows an authenticated user, such as one with the org_admin role, to escalate their privileges and gain full administrative access to the platform. This occurs because the /user/update endpoint, while correctly restricting users to updating only their own account, fails to restrict which specific fields can be modified. As a result, an attacker can change their own user_role to proxy_admin, granting them unauthorized control over all users, teams, keys, models, and prompt history. To address this issue, users should upgrade LiteLLM to version 1.83.10. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-47102.

In SQLite versions before 3.53.2 a high severity vulnerability CVE-2026-11824 was detected. This vulnerability allows an attacker to cause a Denial of Service (crash) or potentially execute arbitrary code. This occurs due to a heap-based buffer overflow in the FTS5 full-text search extension (specifically within the fts5ChunkIterate() function). By supplying a specially crafted database with malicious continuation page metadata (where the szLeaf value is smaller than 4), an attacker can trigger an integer underflow. This results in an inflated remaining byte count during FTS5 MATCH query processing, leading to the overflow of attacker-controlled data into the heap. This vulnerability affects applications compiled with the SQLITE_ENABLE_FTS5 flag. To address this issue, users should upgrade SQLite to version 3.53.2. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-11824.

In Graphite versions before 1.3.15 a high severity vulnerability CVE-2026-50593 was detected. This vulnerability allows an attacker to cause an out-of-bounds memory write, potentially leading to arbitrary code execution, memory corruption, or a denial of service. This occurs due to an integer underflow in the slotat function, which fails to properly validate that an offset is within the allowed slot-map range when processing Graphite actions. To address this issue, users should upgrade Graphite to version 1.3.15. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-50593.

In Milvus versions up to 2.6.13 a medium severity vulnerability CVE-2026-10814 was detected. This vulnerability may allow a local attacker to compromise the system through the exploitation of a weak cryptographic hash. This occurs because the Grantee ID Hash Handler component (specifically in internal/metastore/kv/rootcoord/kv_catalog.go) utilizes a weak hashing algorithm, which an attacker could potentially manipulate, despite the attack complexity being high. To address this issue, users should apply the recommended patch (commit 3d932f1c3e065351c4440c27abe1e6479752544d). For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-10814.

In Weaviate versions up to 1.37.7 a medium severity vulnerability CVE-2026-11500 was detected. This vulnerability allows a remote attacker to bypass intended access controls and gain unauthorized access. This occurs due to improper validation of the StaticApiKey argument within the validateConfig function of the Static API Key Handler (internal/usecases/auth/authentication/apikey/client.go). Although the attack complexity is high and exploitability is considered difficult, a public exploit is available. To address this issue, users should apply the patch (commit 40f2cc32279f0f8a51016c3c6870a2c0c808e6c0) or upgrade Weaviate to version 1.38.0-rc.0. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-11500.

In MLflow versions up to 3.10.0 a low severity vulnerability CVE-2026-10803 was detected. This vulnerability may allow a local attacker to compromise dataset integrity or cause hash collisions. This occurs because the Dataset Digest Computation component (specifically the mlflow.data.digest_utils function in mlflow/data/digest_utils.py) utilizes a weak cryptographic hashing algorithm. Although the attack complexity is rated as high and exploitability is difficult, a proof of concept has been published. There is no fix to this yet. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-10803.

In MLflow versions up to 3.10.1.dev0 a critical severity vulnerability CVE-2026-2651 was detected. This vulnerability allows an attacker to overwrite artifacts belonging to other users, potentially leading to model supply chain poisoning and arbitrary code execution when compromised models are loaded. This occurs because the authorization logic fails to enforce resource-level permission checks for multipart upload (MPU) endpoints (/mlflow-artifacts/mpu/*) when the –serve-artifacts mode is enabled, enabling unauthorized cross-user writes. To address this issue, users should upgrade MLflow to version 3.10.0. For more details, visit https://nvd.nist.gov/vuln/detail/CVE-2026-2651.