Briefs

A critical vulnerability, CVE-2026-48752, in Incus versions prior to 7.2.0 allows an unauthenticated attacker to achieve arbitrary file read and write on the host system via specially crafted container images or instance backups containing unsanitized symlinks, potentially leading to arbitrary command execution as root.

The Incus `incusd` daemon, specifically its S3 protocol multipart upload endpoint in versions prior to 7.1.0, is vulnerable to CVE-2026-48753, a critical path traversal flaw via the `uploadId` parameter, enabling unauthenticated attackers to write arbitrary files to any location on the host system, which can be leveraged for persistent arbitrary command execution.

An argument injection vulnerability (CVE-2026-48755) exists in Incus due to improper validation of the user-provided backup compression algorithm, allowing an authenticated attacker to inject arbitrary arguments into the command line, leading to an arbitrary file write on the host and subsequent arbitrary command execution.

A critical arbitrary file write vulnerability (CVE-2026-48769) exists in the Incus client daemon (`incusd`) when processing images from a malicious server, allowing an attacker to inject path traversal into the `Incus-Image-Hash` header to create arbitrary files in sensitive locations as root, ultimately leading to arbitrary command execution.

An authentication bypass vulnerability in `backpropagate` versions >= 1.1.0 and < 1.2.0 allows unauthenticated attackers to gain full control over the Reflex web UI, even when HTTP Basic authentication is ostensibly enabled via the `--auth` flag, permitting data exfiltration, arbitrary training runs, HuggingFace Hub push, disk-fill DoS, and sensitive path discovery.

A critical privilege escalation vulnerability exists in Rancher Manager where a Project Owner can modify Pod Security Admission (PSA) labels on namespaces within their projects, allowing them to configure a namespace to use the privileged profile, which enables the deployment of privileged workloads that bypass standard container isolation, leading to host-level access and cluster privilege escalation.

Deepstream server versions up to and including 10.0.4 are vulnerable to prototype pollution (CVE-2026-49252), a critical flaw allowing any authenticated user with write permissions to any record to potentially escalate their privileges; the vulnerability is patched in version 10.0.5.

Flowise versions before 2.1.4 are critically vulnerable to configuration injection (CVE-2024-58351) via the `overrideConfig` option in both its frontend web integration and backend Prediction API, which, due to a bypassable `vm2` sandbox, allows attackers to achieve remote code execution, sandbox escape, denial of service, server-side request forgery, prompt injection, and server variable/data exfiltration.

An authentication logic flaw in Cap-go versions prior to 12.128.2 allows attackers to register an account with a victim's unverified email address, then enable two-factor authentication on this pre-registered account to gain full control, read/modify data, enforce organization-level policies, and deny the legitimate user access.

Cap-go versions prior to 12.128.2 are susceptible to an authentication bypass vulnerability (CVE-2026-56073) in OTP verification that allows attackers to manipulate server responses to falsely mark verification successful, leading to unauthorized 2FA enablement and subsequent account takeover.

A critical missing authorization vulnerability, CVE-2026-48582, in Microsoft Exchange Online allows an already authenticated attacker to elevate their privileges over the network, potentially leading to unauthorized access to sensitive data or configuration changes within affected organizations.

A critical improper authentication vulnerability, CVE-2026-45480, in Microsoft Azure Active Directory allows an unauthorized attacker to bypass authentication mechanisms and elevate privileges over a network, potentially leading to full administrative control of Azure AD and associated resources.

An unauthenticated attacker can trigger a denial-of-service condition in applications using the Faraday Ruby library by sending deeply nested query parameters (CVE-2026-54297), leading to `SystemStackError` and application crashes due to uncontrolled recursion.

A stored cross-site scripting (XSS) vulnerability, identified as CVE-2026-54527, in the `jupyterlab-git` JupyterLab extension (versions >= 0.30.0b3, < 0.54.0a1), specifically in `PlainTextDiff.ts`, allows an adversary with Git commit access to execute arbitrary JavaScript in a victim's browser and achieve Remote Code Execution (RCE) on the JupyterLab server by crafting a malicious filename in a Git commit that, when viewed as a rename diff, triggers the XSS payload to steal `_xsrf` cookies, open a terminal, and execute arbitrary shell commands to exfiltrate data.

An authenticated user can bypass the admin-configured `excluded_paths` security control in `jupyterlab-git` versions up to 0.53.0 by exploiting a case-sensitivity flaw on case-insensitive filesystems (e.g., macOS APFS, Windows NTFS), allowing unauthorized read access to git history and file content in explicitly excluded directories.

The `Oj.dump` function in the Ruby `oj` gem, when operating in object mode, is vulnerable to a heap buffer overflow (CVE-2026-54896) when serializing `Exception` objects with an excessively large `:indent` value, leading to memory corruption and potential denial of service or remote code execution.

A heap use-after-free vulnerability (CVE-2026-54897) exists in `Oj::Doc` iterators (`each_value`, `each_child`, `each_leaf`) in the `oj` Ruby gem, allowing an attacker to cause application crashes or unpredictable behavior when a Ruby block yielded during iteration reentrantly calls `doc.close` or `d.close`.

The `Oj.dump` function in the `Oj` Ruby gem is vulnerable to a stack-based buffer overflow (CVE-2026-54502) due to improper validation of the `:indent` parameter, allowing an attacker to trigger a process crash or potentially remote code execution by providing an excessively large integer value, affecting all `Oj` gem versions prior to `3.17.2`.

A critical vulnerability (CVE-2026-53488) exists in the containerd CRI plugin where image configuration `LABEL` instructions are propagated to containers without validation, allowing an attacker to inject and execute arbitrary commands with host-root privileges on the underlying host when a maliciously crafted container image is pulled and processed by specific plugins.

A high-severity vulnerability (CVE-2026-53489) in containerd's CRI plugin allows an unprivileged attacker to read arbitrary files on the host system by crafting a malicious checkpoint with a symlink that `containerd` follows during `container.log` restoration, enabling data exfiltration via `kubectl logs`.

A high-severity vulnerability (CVE-2026-53492) in containerd's CRI implementation allows an attacker with pod creation permissions to smuggle arbitrary Container Device Interface (CDI) annotations during container restoration, bypassing Kubernetes resource allocation and enabling unauthorized device and host mount injection into the restored container.

Stanza, an NLP library, is vulnerable to remote code execution (CVE-2026-54499) due to an unsafe fallback mechanism when loading PyTorch model files, allowing an attacker who can place a malicious pretrain or model file to achieve arbitrary code execution on systems processing NLP pipelines, leading to credential theft, backdoors, data exfiltration, and lateral movement.

An unauthenticated information disclosure vulnerability (CVE-2023-54357) in the Joomla com_booking component version 2.4.9 allows attackers to enumerate user accounts, including names, usernames, and email addresses, by exploiting the getUserData function via specific GET requests.

A Server-Side Request Forgery (SSRF) vulnerability exists in Hugo versions 0.162.0 through 0.163.0, where the 'security.http.urls' policy designed to deny requests to loopback, internal, and cloud-metadata IPv4 literals could be bypassed as the policy only matched dotted-decimal notation, allowing alternate IPv4 encodings (integer, hex, octal) to pass, enabling build-time server-side requests to internal services and cloud-metadata endpoints when untrusted or data-derived URLs are passed to 'resources.GetRemote'.

An unauthenticated API endpoint, `GET /api/pages/nested`, in Alchemy CMS versions up to 8.2.5 (including all 8.x versions prior to a fix and all 7.x versions up to 7.4.14), fails to enforce authorization and scoping checks, allowing any anonymous user to retrieve the complete page tree, encompassing restricted and unpublished pages, and, with `?elements=true`, the full content of these sensitive pages, completely bypassing intended access controls and leading to unauthorized information disclosure.

An unauthenticated attacker can exploit CVE-2017-20267, an SQL injection vulnerability in Joomla! Component Calendar Planner 1.0.1, by sending malicious GET requests to the 'events' view via the 'category_id' parameter, allowing for sensitive database information extraction.

An SQL injection vulnerability, CVE-2017-20266, in Joomla SP Movie Database version 1.3 allows unauthenticated attackers to execute arbitrary SQL queries by injecting malicious code through the `searchword` parameter in GET requests to the `searchresults` view, enabling extraction of sensitive database information.

An SQL injection vulnerability, CVE-2017-20265, in Joomla! Component Flip Wall 8.0 allows unauthenticated attackers to execute arbitrary SQL queries via malicious GET requests to the `wallid` parameter, enabling the extraction of sensitive database information.

An unauthenticated SQL injection vulnerability (CVE-2017-20264) in Joomla! Component Sponsor Wall version 8.0 allows attackers to execute arbitrary SQL queries by injecting malicious code into the `wallid` parameter of GET requests to `index.php`, leading to the extraction of sensitive database information such as credentials and configuration data.

An unauthenticated SQL injection vulnerability (CVE-2017-20263) in Joomla! Component FocalPoint Pro/Free version 1.2.3 allows attackers to execute arbitrary SQL queries via a crafted 'id' parameter in GET requests, leading to sensitive database information disclosure.