Security Feed

The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True. One can create a specially crafted .keras model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special config.json (a file within the .keras archive) that will invoke keras.config.enable_unsafe_deserialization() to disable safe mode. Once safe mode is disable, one can use the Lambda layer feature of keras, w

A vulnerability was found in 07FLYCMS, 07FLY-CMS and 07FlyCRM up to 20250831. This issue affects some unknown processing of the file /index.php/Login/login. Performing manipulation of the argument Username results in sql injection. It is possible to initiate the attack remotely. The exploit has been made public and could be used. This product is published under multiple names. The vendor was contacted early about this disclosure but did not respond in any way.

An issue was discovered in GitLab CE/EE affecting all versions before 18.2.7, 18.3 before 18.3.3, and 18.4 before 18.4.1 that allows unauthenticated users to cause a Denial of Service (DoS) condition while uploading specifically crafted large JSON files.

libvips is a demand-driven, horizontally threaded image processing library. For versions 8.17.1 and below, when libvips is compiled with support for PDF input via poppler, the pdfload operation is affected by a buffer read overflow when parsing the header of a crafted PDF with a page that defines a width but not a height. Those using libvips compiled without support for PDF input are unaffected as well as thosewith support for PDF input via PDFium. This issue is fixed in version 8.17.2. A workar

FreshRSS is a free, self-hostable RSS aggregator. Versions 1.26.3 and below expose information about feeds and tags of default admin users, due to lack of access checking in the FreshRSS_Auth::hasAccess() function used by some of the tag/feed related endpoints. FreshRSS controllers usually have a defined firstAction() method with an override to make sure that every action requires access. If one doesn't, then every action has to check for access manually, and certain endpoints use neither the fi

A local privilege escalation vulnerability exists in the safe_asterisk script included with the Asterisk toolkit package. When Asterisk is started via this script (common in SysV init or FreePBX environments), it sources all .sh files located in /etc/asterisk/startup.d/ as root, without validating ownership or permissions. Non-root users with legitimate write access to /etc/asterisk can exploit this behaviour by placing malicious scripts in the startup.d directory, which will then execute with

AliasVault is a privacy-first password manager with built-in email aliasing. A server-side request forgery (SSRF) vulnerability exists in the favicon extraction feature of AliasVault API versions 0.23.0 and lower. The extractor fetches a user-supplied URL, parses the returned HTML, and follows <link rel="icon" href="…">. Although the initial URL is validated to allow only HTTP/HTTPS with default ports, the extractor automatically follows redirects and does not block requests to loopback or inter

Rack is a modular Ruby web server interface. Prior to version 2.2.18, Rack::QueryParser enforces its params_limit only for parameters separated by &, while still splitting on both & and ;. As a result, attackers could use ; separators to bypass the parameter count limit and submit more parameters than intended. Applications or middleware that directly invoke Rack::QueryParser with its default configuration (no explicit delimiter) could be exposed to increased CPU and memory consumption. This can

Deserialization of Untrusted Data vulnerability in ConveyThis ConveyThis conveythis-translate allows Object Injection.This issue affects ConveyThis: from n/a through <= 269.1.

Authlib is a Python library which builds OAuth and OpenID Connect servers. Prior to version 1.6.4, Authlib’s JWS verification accepts tokens that declare unknown critical header parameters (crit), violating RFC 7515 “must‑understand” semantics. An attacker can craft a signed token with a critical header (for example, bork or cnf) that strict verifiers reject but Authlib accepts. In mixed‑language fleets, this enables split‑brain verification and can lead to policy bypass, replay, or privilege es

Denial of Service issue in GraphQL endpoints in Gitlab EE/CE affecting all versions from 11.10 prior to 18.2.7, 18.3 prior to 18.3.3, and 18.4 prior to 18.4.1 allows unauthenticated users to potentially bypass query complexity limits leading to resource exhaustion and service disruption.

The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True. One can create a specially crafted .h5/.hdf5 model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special .h5 archive file that uses the Lambda layer feature of keras which allows arbitrary Python code in the form of pickled code. The vulnerability comes from the fact that the safe_mode=True option is not h

Use-after-free in the DOM: Core & HTML component. This vulnerability was fixed in Firefox ESR 153.4, Thunderbird 157, Thunderbird 140.17, Thunderbird 153.4, Firefox 157, Firefox ESR 115.42, and Firefox ESR 140.17.

JIT miscompilation in the JavaScript: WebAssembly component. This vulnerability was fixed in Firefox ESR 153.4, Thunderbird 157, Thunderbird 140.17, Thunderbird 153.4, Firefox 157, and Firefox ESR 140.17.

Use-after-free in the JavaScript: WebAssembly component. This vulnerability was fixed in Thunderbird 157 and Firefox 157.

Privilege escalation due to use-after-free in the Graphics: WebRender component. This vulnerability was fixed in Firefox ESR 153.4, Thunderbird 157, Thunderbird 140.17, Thunderbird 153.4, Firefox 157, Firefox ESR 115.42, and Firefox ESR 140.17.

Arcane is an interface for managing Docker containers, images, networks, and volumes. Prior to 2.6.0, UpdateUser in backend/api/handlers/users.go sets PasswordHash for any target without comparing caller and target privilege. A holder of delegated users:update permission can submit a password change for the global administrator, replace the administrator password, and authenticate as that account, resulting in complete global administrator account takeover. This issue is fixed in version 2.6.0.

A flaw was found in GIMP. An integer overflow occurs in the file-pvr plug-in when calculating memory buffer sizes for image dimensions. By tricking a user into opening or previewing a specially crafted PowerVR (.pvr) file, an attacker can trigger a heap-based buffer overflow, potentially leading to arbitrary code execution or a Denial of Service (DoS) via an application crash.