| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| The modelscope/ms-swift library thru 2.6.1 is vulnerable to arbitrary code execution through deserialization of untrusted data within the `load_model_meta()` function of the `ModelFileSystemCache()` class. Attackers can execute arbitrary code and commands by crafting a malicious serialized `.mdl` payload, exploiting the use of `pickle.load()` on data from potentially untrusted sources. This vulnerability allows for remote code execution (RCE) by deceiving victims into loading a seemingly harmless checkpoint during a normal training process, thereby enabling attackers to execute arbitrary code on the targeted machine. Note that the payload file is a hidden file, making it difficult for the victim to detect tampering. More importantly, during the model training process, after the `.mdl` file is loaded and executes arbitrary code, the normal training process remains unaffected'meaning the user remains unaware of the arbitrary code execution. |
| Himmelblau is an interoperability suite for Microsoft Azure Entra ID and Intune. When debugging is enabled for Himmelblau in version 1.0.0, the himmelblaud_tasks service leaks an Intune service access token to the system journal. This short-lived token can be used to detect the host's Intune compliance status, and may permit additional administrative operations for the Intune host device (though the API for these operations is undocumented). This is fixed in version 1.1.0. To workaround this issue, ensure that Himmelblau debugging is disabled. |
| GRAU DATA Blocky before 3.1 stores passwords encrypted rather than hashed. At the login screen, the user's password is compared to the user's decrypted cleartext password. An attacker with Windows admin or debugging rights can therefore steal the user's Blocky password and from there impersonate that local user. |
| In affected versions of the Octopus Kubernetes worker or agent, sensitive variables could be written to the Kubernetes script pod log in clear-text. This was identified in Version 2 however it was determined that this could also be achieved in Version 1 and the fix was applied to both versions accordingly. |
| Hugging Face Diffusers CogView4 Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Hugging Face Diffusers. 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 checkpoints. 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 process. Was ZDI-CAN-27424. |
| Deserialization of Untrusted Data vulnerability in eyecix JobSearch wp-jobsearch.This issue affects JobSearch: from n/a through < 3.0.8. |
| CCleaner v5.33.6162 and CCleaner Cloud v1.07.3191 (32-bit builds) contained a malicious pre-entry-point loader that diverts execution from __scrt_common_main_seh into a custom loader. That loader decodes an embedded blob into shellcode, allocates executable heap memory, resolves Windows API functions at runtime, and transfers execution to an in-memory payload. The payload performs anti-analysis checks, gathers host telemetry, encodes the data with a two-stage obfuscation, and attempts HTTPS exfiltration to hard-coded C2 servers or month-based DGA domains. Potential impacts include remote data collection and exfiltration, stealthy in-memory execution and persistence, and potential lateral movement. CCleaner was developed by Piriform, which was acquired by Avast in July 2017; Avast later merged with NortonLifeLock to form the parent company now known as Gen Digital. According to vendor advisories, the compromised CCleaner build was released on August 15, 2017 and remediated on September 12, 2017 with v5.34; the compromised CCleaner Cloud build was released on August 24, 2017 and remediated on September 15, 2017 with v1.07.3214. |
| Web Developer for Chrome v0.4.9 contained malicious code that generated a domain via a DGA and fetched a remote script. The fetched script conditionally loaded follow-on modules that performed extensive ad substitution and malvertising, displayed fake “repair” alerts that redirected users to affiliate programs, and attempted to harvest credentials when users logged in. Injected components enumerate common banner sizes for substitution, replace third-party ad calls, and redirect victim traffic to affiliate landing pages. Potential impacts include user-level code execution in the browser context, large-scale ad fraud and traffic hijacking, credential theft, and exposure to additional payloads delivered by the actor. The compromise was reported on by the maintainer of Web Developer for Chrome on August 2, 2017 and remediated in v0.5.0. |
| VestaCP commit a3f0fa1 (2018-05-31) up to commit ee03eff (2018-06-13) contain embedded malicious code that resulted in a supply-chain compromise. New installations created from the compromised installer since at least May 2018 were subject to installation of Linux/ChachaDDoS, a multi-stage DDoS bot that uses Lua for second- and third-stage components. The compromise leaked administrative credentials (base64-encoded admin password and server domain) to an external URL during installation and/or resulted in the installer dropping and executing a DDoS malware payload under local system privileges. Compromised servers were subsequently observed participating in large-scale DDoS activity. Vesta acknowledged exploitation in the wild in October 2018. |
| Prebid Universal Creative (PUC) is a JavaScript API to render multiple formats. Npm users of PUC 1.17.3 or PUC latest were briefly affected by crypto-related malware. This includes the extremely popular jsdelivr hosting of this file. The maintainers of PUC unpublished version 1.17.3. Users should see Prebid.js 9 release notes for suggestions on moving off the deprecated workflow of using the PUC or pointing to a dynamic version of it. PUC users pointing to latest should transition to 1.17.2 as soon as possible to avoid similar attacks in the future. |
| The OVRI Payment plugin for WordPress contains malicious .htaccess files in version 1.7.0. The files contain directives to prevent the execution of certain scripts while allowing execution of known malicious PHP files. If moved outside of the plugin's directory, they may interfere with the proper function of a site. |
| Insufficiently Protected Credentials vulnerability in OpenText™ AccuRev allows Authentication Bypass. When installed on a Linux or Solaris system
the vulnerability could allow anyone who knows a valid AccuRev username can use the AccuRev client to login and gain access to AccuRev source control without knowing the user’s password.
This issue affects AccuRev: 2017.1. |
| Ubee EVW3226 cable modem/routers firmware versions up to and including 1.0.20 store configuration backup files in the web root after they are generated for download. These backup files remain accessible without authentication until the next reboot. A remote attacker on the local network can request 'Configuration_file.cfg' directly to obtain the backup archive. Because backup files are not encrypted, they expose sensitive information including the plaintext admin password, allowing full compromise of the device. |
| Insufficiently Protected Credentials in the Crowdstrike connector can lead to Crowdstrike credentials being leaked. A malicious user can access cached credentials from a Crowdstrike connector in another space by creating and running a Crowdstrike connector in a space to which they have access. |
| Three Bitnami Helm charts mount Kubernetes Secrets under a predictable path (/opt/bitnami/*/secrets) that is located within the web server document root.
In affected versions, this can lead to unauthenticated access to sensitive credentials via HTTP/S. A remote attacker could retrieve these secrets by accessing specific URLs if the application is exposed externally. The issue affects deployments using the default value of usePasswordFiles=true, which mounts secrets as files into the container filesystem. |
| Deserialization of Untrusted Data vulnerability in Hernan Villanueva Boldermail boldermail allows Object Injection.This issue affects Boldermail: from n/a through <= 2.4.0. |
| Hugging Face smolagents Remote Python Executor Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Hugging Face smolagents. Authentication is not required to exploit this vulnerability.
The specific flaw exists within the parsing of pickle data. 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-28312. |
| Bluetooth HCI Adaptor from Realtek has a Link Following vulnerability. Local attackers with regular privileges can create a symbolic link with the same name as a specific file, causing the product to delete arbitrary files pointed to by the link. Subsequently, attackers can leverage arbitrary file deletion to privilege escalation. |
| In Splunk Add-on for Palo Alto Networks versions below 2.0.2, the add-on exposes client secrets in plain text in the _internal index during the addition of new “Data Security Accounts“. The vulnerability would require either local access to the log files or administrative access to internal indexes, which by default only the admin role receives. Review roles and capabilities on your instance and restrict internal index access to administrator-level roles. See [Define roles on the Splunk platform with capabilities](https://docs.splunk.com/Documentation/Splunk/latest/Security/Rolesandcapabilities) in the Splunk documentation for more information. |
| Medtronic MyCareLink Patient Monitor has an internal service that deserializes data, which allows a local attacker to interact with the service by crafting a binary payload to crash the service or elevate privileges.
This issue affects MyCareLink Patient Monitor models 24950 and 24952: before June 25, 2025 |
