The CosyVoice project thru commit 6e01309e01bc93bbeb83bdd996b1182a81aaf11e (2025-30-21) contains an insecure deserialization vulnerability (CWE-502) in its model loading process. When loading model files (.pt) from a user-specified directory (via the --model_dir argument), the code uses torch.load() without the security-restrictive weights_only=True parameter. This allows the deserialization of arbitrary Python objects via the Pickle module. An attacker can exploit this by providing a maliciously crafted model directory containing .pt files with embedded pickle payloads. When a victim loads this directory using CosyVoice's web interface, the malicious payload is executed, leading to remote code execution on the victim's system.

Apache Karaf: Decanter log-socket collector has deserialization vulnerability

Deserialization of Untrusted Data vulnerability in Apache Karaf Decanter. The Decanter log socket collector exposes the port 4560, without authentication. If the collector exposes allowed classes property, this configuration can be bypassed. It means that the log socket collector is vulnerable to deserialization of untrusted data, eventually causing DoS. NB: Decanter log socket collector is not installed by default. Users who have not installed Decanter log socket are not impacted by this issue. This issue affects Apache Karaf Decanter before 2.12.0. Users are recommended to upgrade to version 2.12.0, which fixes the issue.

vLLM allows a malicious model RCE by torch.load in hf_model_weights_iterator

vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling. This vulnerability is fixed in v0.7.0.

WordPress Asgaros Forum Plugin <= 2.7.2 is vulnerable to PHP Object Injection

Deserialization of Untrusted Data vulnerability in Thomas Belser Asgaros Forum.This issue affects Asgaros Forum: from n/a through 2.7.2.

Remote Code Execution via Model Deserialization in invoke-ai/invokeai

A remote code execution vulnerability exists in invoke-ai/invokeai versions 5.3.1 through 5.4.2 via the /api/v2/models/install API. The vulnerability arises from unsafe deserialization of model files using torch.load without proper validation. Attackers can exploit this by embedding malicious code in model files, which is executed upon loading. This issue is fixed in version 5.4.3.

WP Simple Galleries <= 1.34 - Authenticated (Contributor+) PHP Object Injection

The WP Simple Galleries plugin for WordPress is vulnerable to PHP Object Injection in versions up to, and including, 1.34 via deserialization of untrusted input from the 'wpsimplegallery_gallery' post meta via 'wpsgallery' shortcode. This allows authenticated attackers, with contributor-level permissions and above, to inject a PHP Object. No POP chain is present in the vulnerable plugin. If a POP chain is present via an additional plugin or theme installed on the target system, it could allow the attacker to delete arbitrary files, retrieve sensitive data, or execute code.

Weaver Xtreme Theme Support < 6.3.1 - Admin+ PHP Object Injection

The Weaver Xtreme Theme Support WordPress plugin before 6.3.1 unserialises the content of an imported file, which could lead to PHP object injections issues when a high privilege user import a malicious file and a suitable gadget chain is present on the blog.

JSCAPE MFT Server Unsafe Deserialization on Management Port

Unsafe deserialization in JSCAPE MFT Server versions prior to 2023.1.9 (Windows, Linux, and MacOS) permits an attacker to run arbitrary Java code (including OS commands) via its management interface

Deserialization of untrusted data in MelisFront

MelisFront is the engine that displays website hosted on Melis Platform. It deals with showing pages, plugins, URL rewritting, search optimization and SEO, etc. Attackers can deserialize arbitrary data on affected versions of `melisplatform/melis-front`, and ultimately leads to the execution of arbitrary PHP code on the system. Conducting this attack does not require authentication. Users should immediately upgrade to `melisplatform/melis-front` >= 5.0.1. This issue was addressed by restricting allowed classes when deserializing user-controlled data.

Deserialization of untrusted data in MelisCms

MelisCms provides a full CMS for Melis Platform, including templating system, drag'n'drop of plugins, SEO and many administration tools. Attackers can deserialize arbitrary data on affected versions of `melisplatform/melis-cms`, and ultimately leads to the execution of arbitrary PHP code on the system. Conducting this attack does not require authentication. Users should immediately upgrade to `melisplatform/melis-cms` >= 5.0.1. This issue was addressed by restricting allowed classes when deserializing user-controlled data.

This vulnerability allows remote attackers to execute arbitrary code on affected installations of Ivanti Avalanche 6.3.2.3490. Although authentication is required to exploit this vulnerability, the existing authentication mechanism can be bypassed. The specific flaw exists within the Notification Server service. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-15448.

SolarWinds Platform Deserialization of Untrusted Data

SolarWinds Platform was susceptible to the Deserialization of Untrusted Data. This vulnerability allows a remote adversary with Orion admin-level account access to SolarWinds Web Console to execute arbitrary commands.

Laravel deserialization

A vulnerability, which was classified as critical, was found in Laravel 5.1. Affected is an unknown function. The manipulation leads to deserialization. It is possible to launch the attack remotely. The exploit has been disclosed to the public and may be used. The identifier of this vulnerability is VDB-206688.

laravel deserialization

A vulnerability was found in laravel 5.1 and classified as problematic. This issue affects some unknown processing. The manipulation leads to deserialization. The attack may be initiated remotely. The exploit has been disclosed to the public and may be used. The identifier VDB-206501 was assigned to this vulnerability.

LiteDB contains Deserialization of Untrusted Data

LiteDB is a small, fast and lightweight .NET NoSQL embedded database. Versions prior to 5.0.13 are subject to Deserialization of Untrusted Data. LiteDB uses a special field in JSON documents to cast different types from `BsonDocument` to POCO classes. When instances of an object are not the same of class, `BsonMapper` use a special field `_type` string info with full class name with assembly to be loaded and fit into your model. If your end-user can send to your app a plain JSON string, deserialization can load an unsafe object to fit into your model. This issue is patched in version 5.0.13 with some basic fixes to avoid this, but is not 100% guaranteed when using `Object` type. The next major version will contain an allow-list to select what kind of Assembly can be loaded. Workarounds are detailed in the vendor advisory.

A php unserialize vulnerability exists in the Ai-Bolit functionality of CloudLinux Inc Imunify360 5.10.2. A specially-crafted malformed file can lead to potential arbitrary command execution. An attacker can provide a malicious file to trigger this vulnerability.

LangGraph Checkpoint: Unsafe JSON deserialization in checkpoint loading

LangGraph SQLite Checkpoint is an implementation of LangGraph CheckpointSaver that uses SQLite DB (both sync and async, via aiosqlite). In versions 4.1.0 and prior, the JsonPlusSerializer can reconstruct Python objects from JSON checkpoint payloads. Under conditions where someone could modify checkpoint bytes at rest in the backing store, the deserialization path could reconstruct objects beyond what the application expects, which could in turn result in code execution at checkpoint load time. This is a defense-in-depth issue. The affected behavior is reachable only when checkpoint bytes at rest in the backing store can be modified by an unauthorized party. In most deployments that prerequisite already implies a serious incident; the additional concern is turning "checkpoint-store write access" into code execution in the application runtime. This issue has been fixed in version 4.1.1.

openstack-ironic: OpenStack Ironic: Denial of Service via crafted JSON string

In OpenStack Ironic 32 before 37.0.0, an unauthenticated malicious user could submit a crafted JSON string to some endpoints on the API or JSON-RPC service and effect a service crash.

WordPress Modula Image Gallery plugin <= 2.14.18 - PHP Object Injection vulnerability

Author PHP Object Injection in Modula Image Gallery <= 2.14.18 versions.

WordPress Post Duplicator plugin <= 3.0.10 - PHP Object Injection vulnerability

Contributor PHP Object Injection in Post Duplicator <= 3.0.10 versions.