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.

The Drag and Drop Multiple File Upload for Contact Form 7 plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 1.3.8.7 via deserialization of untrusted input from the 'dnd_upload_cf7_upload' function. This makes it possible for attackers to inject a PHP Object through a PHAR file. No known POP chain is present in the vulnerable software, which means this vulnerability has no impact unless another plugin or theme containing a POP chain is installed on the site. If a POP chain is present via an additional plugin or theme installed on the target system, it may allow the attacker to perform actions like delete arbitrary files, retrieve sensitive data, or execute code depending on the POP chain present. This vulnerability may be exploited by unauthenticated attackers when a form is present on the site with the file upload action. The Flamingo plugin must be installed and activated in order to exploit the vulnerability. The vulnerability was partially patched in version 1.3.8.8.

Delta Electronics InfraSuite Device Master versions 00.00.01a and prior deserialize network packets without proper verification. If the device connects to an attacker-controlled server, the attacker could send maliciously crafted packets that would be deserialized and executed, leading to remote code execution.

Ray prior to 2.56.0 contains an unsafe deserialization vulnerability in the WebDataset reader that allows attackers to achieve remote code execution by supplying a malicious tar archive to the read_webdataset() function. The _default_decoder() function in webdataset_datasource.py unconditionally calls pickle.loads() on tar entries with .pkl/.pickle extensions and torch.load() with weights_only=False on .pt/.pth entries, executing arbitrary code inside Ray remote workers on every worker that processes the malicious archive.

Trail of Bits fickling versions up to and including 0.1.10 do not include the Python standard library modules _posixsubprocess, site, and atexit in the UNSAFE_IMPORTS denylist (fickle.py). Because these modules are absent from the denylist, fickling's check_safety() function returns LIKELY_SAFE with zero findings for pickle payloads that invoke dangerous functions including _posixsubprocess.fork_exec (C-level process spawner capable of executing arbitrary binaries), site.execsitecustomize (executes arbitrary site customization code), and atexit._run_exitfuncs (triggers all registered exit handler callbacks). The fickling.load() API chains check_safety() into pickle.loads() as an explicit security gate; a LIKELY_SAFE verdict causes the payload to be deserialized and executed. This shares the same root cause as CVE-2026-22607 (cProfile), CVE-2025-67748 (pty), and CVE-2025-67747 (marshal/types). OvertlyBadEvals does not flag these modules because they are standard library imports. UnsafeImports does not flag them because they are not in the denylist. The UnusedVariables heuristic is defeated by the SETITEMS opcode pattern.

The Shortcodes and extra features for Phlox theme WordPress plugin before 2.10.7 unserializes the content of an imported file, which could lead to PHP object injection when a user imports (intentionally or not) a malicious file and a suitable gadget chain is present on the blog.

The Jetpack CRM plugin for WordPress is vulnerable to PHAR deserialization via the ‘zbscrmcsvimpf’ parameter in the 'zeroBSCRM_CSVImporterLitehtml_app' function in versions up to, and including, 5.3.1. While the function performs a nonce check, steps 2 and 3 of the check do not take any action upon a failed check. These steps then perform a 'file_exists' check on the value of 'zbscrmcsvimpf'. If a phar:// archive is supplied, its contents will be deserialized and an object injected in the execution stream. This allows an unauthenticated attacker to obtain object injection if they are able to upload a phar archive (for instance if the site supports image uploads) and then trick an administrator into performing an action, such as clicking a link.

opensearch-ruby is a community-driven, open source fork of elasticsearch-ruby. In versions prior to 2.0.1 the ruby `YAML.load` function was used instead of `YAML.safe_load`. As a result opensearch-ruby 2.0.0 and prior can lead to unsafe deserialization using YAML.load if the response is of type YAML. An attacker must be in control of an opensearch server and convince the victim to connect to it in order to exploit this vulnerability. The problem has been patched in opensearch-ruby gem version 2.0.1. Users are advised to upgrade. There are no known workarounds for this issue.

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.

The String Locator plugin for WordPress is vulnerable to deserialization of untrusted input via the 'string-locator-path' parameter in versions up to, and including 2.5.0. This makes it possible for unauthenticated users to call files using a PHAR wrapper, granted they can trick a site administrator into performing an action such as clicking on a link, that will deserialize and call arbitrary PHP Objects that can be used to perform a variety of malicious actions granted a POP chain is also present. It also requires that the attacker is successful in uploading a file with the serialized payload.

The WordPress Infinite Scroll – Ajax Load More plugin for WordPress is vulnerable to deserialization of untrusted input via the 'alm_repeaters_export' parameter in versions up to, and including 5.5.3. This makes it possible for unauthenticated users to call files using a PHAR wrapper, granted they can trick a site administrator into performing an action such as clicking on a link, that will deserialize and call arbitrary PHP Objects that can be used to perform a variety of malicious actions granted a POP chain is also present. It also requires that the attacker is successful in uploading a file with the serialized payload.

The Redirection for Contact Form 7 plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 3.2.4 via deserialization of untrusted input in the get_lead_fields function. This makes it possible for unauthenticated attackers to inject a PHP Object. The additional presence of a POP chain in a Contact Form 7 plugin allows attackers to delete arbitrary files. Additionally, in certain server configurations, Remote Code Execution is possible

An issue was discovered in Public Knowledge Project (PKP) pkp-lib before 3.1.2-2, as used in Open Journal Systems (OJS) before 3.1.2-2. Code injection can occur in the OJS report generator if an authenticated Journal Manager user visits a crafted URL, because unserialize is used.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 0.9.0 or newer, enabling a maliciously uploaded PyFunc model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.1.0 or newer, enabling a maliciously uploaded scikit-learn model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 0.5.0 or newer, enabling a maliciously uploaded PyTorch model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.23.0 or newer, enabling a maliciously uploaded LightGBM scikit-learn model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.1.0 or newer, enabling a maliciously uploaded scikit-learn model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.27.0 or newer, enabling a maliciously crafted Recipe to execute arbitrary code on an end user’s system when run.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 2.0.0rc0 or newer, enabling a maliciously uploaded Tensorflow model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.24.0 or newer, enabling a maliciously uploaded pmdarima model to run arbitrary code on an end user’s system when interacted with.

Deserialization of untrusted data can occur in versions of the MLflow platform running version 2.5.0 or newer, enabling a maliciously uploaded Langchain AgentExecutor model to run arbitrary code on an end user’s system when interacted with.

Insecure deserialization of untrusted data in Micro Focus Service Manager product versions 9.30, 9.31, 9.32, 9.33, 9.34, 9.35, 9.40, 9.41, 9.50, 9.51, 9.52, 9.60, 9.61, 9.62. The vulnerability could be exploited to allow insecure deserialization of untrusted data.

Deserialization of untrusted data can occur in the R statistical programming language, on any version starting at 1.4.0 up to and not including 4.4.0, enabling a maliciously crafted RDS (R Data Serialization) formatted file or R package to run arbitrary code on an end user’s system when interacted with.

Deserialization of Untrusted Data vulnerability in webdevstudios Constant Contact for WordPress allows Object Injection. This issue affects Constant Contact for WordPress: from n/a through 4.1.1.

Deserialization of untrusted data can occur in versions 0.17.0 to 1.14.2 of the client SDK of Allegro AI’s ClearML platform, enabling a maliciously uploaded artifact to run arbitrary code on an end user’s system when interacted with.

The UpdraftPlus: WP Backup & Migration Plugin plugin for WordPress is vulnerable to PHP Object Injection in all versions from 1.23.8 to 1.24.11 via deserialization of untrusted input in the 'recursive_unserialized_replace' function. This makes it possible for unauthenticated attackers to inject a PHP Object. No known POP chain is present in the vulnerable software, which means this vulnerability has no impact unless another plugin or theme containing a POP chain is installed on the site. If a POP chain is present via an additional plugin or theme installed on the target system, it may allow the attacker to perform actions like delete arbitrary files, retrieve sensitive data, or execute code depending on the POP chain present. An administrator must perform a search and replace action to trigger the exploit.

Ghidra before 12.1 contains an unsafe deserialization vulnerability in client-side Shared-Project RMI connection code that allows unauthenticated remote code execution. Attackers can craft a malicious project file with a ghidra:// URL that, when opened via File → Open Project, deserializes untrusted objects using a Jython 2.7.4 gadget chain to execute arbitrary commands.

Hugging Face Transformers Trax Model Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Hugging Face Transformers. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the handling of model files. 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 current user. Was ZDI-CAN-25012.

Hugging Face Transformers MobileViTV2 Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Hugging Face Transformers. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the handling of configuration files. 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 current user. Was ZDI-CAN-24322.

The Migration, Backup, Staging – WPvivid plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 0.9.107 via deserialization of untrusted input in the 'replace_row_data' and 'replace_serialize_data' functions. This makes it possible for unauthenticated attackers to inject a PHP Object. No known POP chain is present in the vulnerable software. 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. An administrator must create a staging site to trigger the exploit.

The Backup Migration plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 1.4.6 via deserialization of untrusted input in the 'recursive_unserialize_replace' function. This makes it possible for unauthenticated attackers to inject a PHP Object. The additional presence of a POP chain allows attackers to delete arbitrary files, retrieve sensitive data, or execute code. An administrator must create a staging site in order to trigger the exploit.

The String locator plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 2.6.6 via deserialization of untrusted input in the 'recursive_unserialize_replace' function. This makes it possible for unauthenticated attackers to inject a PHP Object. No known POP chain is present in the vulnerable software. 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. An administrator must perform a search and replace action to trigger the exploit.

A vulnerability classified as critical was found in van_der_Schaar LAB TemporAI 0.0.3. Affected by this vulnerability is the function load_from_file of the component PKL File Handler. The manipulation leads to deserialization. The attack can be launched remotely. The exploit has been disclosed to the public and may be used. The identifier VDB-252181 was assigned to this vulnerability. NOTE: The vendor was contacted early and confirmed immediately the existence of the issue. A patch is planned to be released in February 2024.

Remote Cross-site Request forgery (CSRF) potential has been identified in UCMBD Browser version 4.10, 4.11, 4.12, 4.13, 4.14, 4.15, 4.15.1 which could allow for remote unsafe deserialization and cross-site request forgery (CSRF).

Remote Cross-site Request forgery (CSRF) potential has been identified in UCMBD Server version DDM Content Pack V 10.20, 10.21, 10.22, 10.22 CUP7, 10.30, 10.31, 10.32, 10.33, 10.33 CUP2, 11.0 and CMS Server version 2018.05 BACKGROUND which could allow for remote unsafe deserialization and cross-site request forgery (CSRF).

Welcart e-Commerce 2.11.6 and earlier versions contains an untrusted data deserialization vulnerability. If this vulnerability is exploited, arbitrary code may be executed by a remote unauthenticated attacker who can access websites created using the product.

Inductive Automation Ignition ResponseParser Notification Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Inductive Automation Ignition. User interaction is required to exploit this vulnerability in that the target must connect to a malicious server. The specific flaw exists within the ResponseParser method. 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 current user. Was ZDI-CAN-22067.

Inductive Automation Ignition ResponseParser SerializedResponse Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Inductive Automation Ignition. User interaction is required to exploit this vulnerability in that the target must connect to a malicious server. The specific flaw exists within the ResponseParser method. 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 current user. Was ZDI-CAN-21926.

Ray is an AI compute engine. From version 2.54.0 to before version 2.55.0, Ray Data registers custom Arrow extension types (ray.data.arrow_tensor, ray.data.arrow_tensor_v2, ray.data.arrow_variable_shaped_tensor) globally in PyArrow. When PyArrow reads a Parquet file containing one of these extension types, it calls __arrow_ext_deserialize__ on the field's metadata bytes. Ray's implementation passes these bytes directly to cloudpickle.loads(), achieving arbitrary code execution during schema parsing, before any row data is read. This issue has been patched in version 2.55.0.

PHPMailer before 5.2.27 and 6.x before 6.0.6 is vulnerable to an object injection attack.

Hugging Face Transformers MaskFormer Model Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Hugging Face Transformers. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the parsing of model files. 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 current user. Was ZDI-CAN-25191.

Deserialization of Untrusted Data vulnerability in BoldGrid weForms weforms allows Object Injection.This issue affects weForms: from n/a through <= 1.6.26.

An issue was discovered in Spipu HTML2PDF before 5.2.4. Attackers can trigger deserialization of arbitrary data via the injection of a malicious <link> tag in the converted HTML document.

Cross-Site Request Forgery (CSRF), Deserialization of Untrusted Data vulnerability in ThemeHigh Job Manager & Career – Manage job board listings, and recruitments.This issue affects Job Manager & Career – Manage job board listings, and recruitments: from n/a through 1.4.4.

PyTorch-Lightning versions 2.6.0 and earlier contain an insecure deserialization vulnerability (CWE-502) in the checkpoint loading mechanism. The LightningModule.load_from_checkpoint() method, which is commonly used to load saved model states, internally calls torch.load() without setting the security-restrictive weights_only=True parameter. This default behavior allows the deserialization of arbitrary Python objects via the Pickle module. A remote attacker can exploit this by providing a maliciously crafted checkpoint file, leading to arbitrary code execution on the victim's system when the file is loaded.

The snorkel library thru v0.10.0 contains an insecure deserialization vulnerability (CWE-502) in the Trainer.load() method of the Trainer class. The method loads model checkpoint files using torch.load() without enabling the security-restrictive weights_only=True parameter. This default behavior allows the deserialization of arbitrary Python objects via the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution on the victim's system when the file is loaded via the vulnerable method.

The snorkel library thru v0.10.0 contains a critical insecure deserialization vulnerability (CWE-502) in the BaseLabeler.load() method of the BaseLabeler class. The method loads serialized labeler models using the unsafe pickle.load() function on user-supplied file paths without any validation or security controls. Python's pickle module is inherently dangerous for deserializing untrusted data, as it can execute arbitrary code during the deserialization process. A remote attacker can exploit this by providing a maliciously crafted pickle file, leading to arbitrary code execution on the victim's system when the file is loaded via the vulnerable method.

The snorkel library thru v0.10.0 contains an insecure deserialization vulnerability (CWE-502) in the MultitaskClassifier.load() method of the MultitaskClassifier class. The method loads model weight files using torch.load() without enabling the security-restrictive weights_only=True parameter. This default behavior allows the deserialization of arbitrary Python objects via the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution on the victim's system when the file is loaded via the vulnerable method.

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.