By year

Vulnerabilities disclosed in 2026

CVEs published in 2026 with SEC.co analysis.

8541 published vulnerabilities · page 9 of 86

Known exploited (KEV)Critical High Medium All recent

CVE-2026-47759HIGH 8.7
TinyMCE, a widely-used open-source rich text editor, contains a stored cross-site scripting (XSS) vulnerability in versions before 5.11.1, 7.9.3, and 8.5.1. Authenticated attackers can craft malicious content that exploits unsanitized data-mce-* attributes (such as data-mce-href, data-mce-src, and data-mce-style) to inject and persist malicious scripts. When the editor serializes the content, attackers' payloads override legitimate attribute protections, allowing arbitrary JavaScript execution in users' browsers. The vulnerability requires user interaction and authentication but affects all users viewing the injected content.
CVE-2026-47760HIGH 8.7
TinyMCE, a widely used open-source rich text editor, contains a cross-site scripting (XSS) vulnerability in versions 6.8.0 through 7.0.x. The flaw stems from improper handling of SVG namespace scope within the editor's HTML sanitizer. An attacker with login access can craft malicious SVG payloads using nested elements that evade the sanitizer's attribute filtering, allowing them to inject and execute arbitrary JavaScript in the context of the application. The vulnerability requires user interaction (clicking or otherwise engaging with the crafted content) but can affect all users viewing that content once injected. This issue is resolved in TinyMCE 7.1.0.
CVE-2026-47761HIGH 8.7
TinyMCE, a widely-used open source rich text editor, contains a stored cross-site scripting (XSS) vulnerability in its media plugin. An authenticated attacker can inject malicious JavaScript by crafting specially formed data-mce-* HTML attributes within media content. When that content is later viewed or rendered by any user, the injected script executes in their browser with the privileges of the application. This is particularly dangerous because the malicious payload persists in the editor's content—it isn't just a temporary injection. Organizations using TinyMCE versions prior to 5.11.1, 7.9.3, or 8.5.1 with the media plugin active should prioritize upgrading immediately.
CVE-2026-47762HIGH 8.7
TinyMCE, a widely-used open-source rich text editor, contains a stored cross-site scripting (XSS) vulnerability that allows authenticated attackers to inject malicious scripts into protected content. The flaw exists in how the editor processes specially crafted comments marked with the mce:protected tag, bypassing built-in sanitization controls. When a user later restores or views the protected content, the injected script executes in their browser. This affects users who have enabled TinyMCE's protect option, a feature designed to mark content as requiring special handling. The vulnerability has been patched in versions 5.11.1, 7.9.3, and 8.5.1.
CVE-2026-48527HIGH 8.7
HAX CMS, a platform for managing distributed microsites with PHP or Node.js backends, contains a stored cross-site scripting (XSS) flaw in its page-editing API. Authenticated users with page-editing permissions can inject malicious scripts that persist in the system by circumventing the HTML sanitizer through a specific formatting bypass. Once stored, these scripts execute in the browsers of other users who view the affected pages, potentially stealing credentials, hijacking sessions, or defacing content.
CVE-2026-48716HIGH 8.7
A vulnerability exists in nanobot versions 0.1.5.post3 and earlier that allows an attacker to write files anywhere on a system where the WhatsApp bridge is running. When a user receives a document via WhatsApp through nanobot's bridge, the application downloads the file and saves it to disk. An attacker can craft a malicious filename containing path traversal sequences (like `../../../`) to escape the intended media directory and write to sensitive locations, such as SSH configuration directories or system files. Since the attacker controls both the filename and file content, they can potentially overwrite critical system or user files. This requires network access and some knowledge of the target system's directory structure, but does not require user interaction beyond receiving a WhatsApp message.
CVE-2026-49368HIGH 8.7
JetBrains YouTrack versions before 2026.1.13162 contain a stored cross-site scripting (XSS) vulnerability in project notification templates. An authenticated attacker can inject malicious scripts into notification templates that will execute in the browsers of other users who view those notifications. This is a stored XSS rather than a reflected attack, meaning the payload persists and affects multiple users over time, increasing its impact potential.
CVE-2026-52805HIGH 8.7
Gogs, a self-hosted Git service, contains a Server-Side Request Forgery (SSRF) vulnerability in its repository migration feature that allows authenticated users to bypass internal network restrictions. When migrating a repository, an attacker can submit a URL that initially appears benign but redirects to an internal system (such as the Gogs server itself or other local services). The application validates the original URL but fails to re-validate after HTTP redirects, allowing git clone to follow the redirect and access restricted internal repositories. An authenticated attacker can then pull the contents of that internal repository into an externally controlled repository, effectively exfiltrating data from systems that should be network-isolated. This flaw affects Gogs versions prior to 0.14.3.
CVE-2026-53230HIGH 8.7
A memory safety flaw exists in the Linux kernel's Mellanox (mlx5) network driver. When querying MAC address lists from a virtual function (VF) with custom network configuration, the driver allocates a buffer that is too small, causing the firmware response to overflow the buffer boundary. This can lead to kernel crashes or potentially allow local privilege escalation. The fix involves reading the correct capacity limits from each virtual function's own configuration rather than assuming all functions match the physical function's limits.
CVE-2026-53608HIGH 8.7
ApostropheCMS, a popular open-source Node.js content management system, contains a stored cross-site scripting (XSS) vulnerability in its SEO configuration module. Editors and content managers can inject malicious code into Google Analytics and Google Tag Manager ID fields, which are then executed in the browsers of all website visitors. Because these fields accept unsanitized input and are rendered directly into page scripts without validation, an attacker with editor privileges can compromise every visitor's session and steal sensitive data.
CVE-2026-54011HIGH 8.7
Open WebUI, a self-hosted AI platform, contains a cross-site scripting (XSS) vulnerability in its Markdown preview feature. When users preview files containing Mermaid diagrams, the application unsafely renders and injects the generated SVG directly into the browser's DOM using innerHTML. Because Mermaid's security level is set to 'loose', an attacker can embed malicious JavaScript within a Mermaid diagram that executes in the victim's browser with the privileges of the Open WebUI application. An authenticated user must view a malicious Markdown file for the attack to succeed. This vulnerability is resolved in version 0.9.6.
CVE-2026-54406HIGH 8.7
A path traversal vulnerability in self-hosted UniFi Network Application instances allows a high-privilege network-adjacent attacker to write files on the underlying host system, potentially leading to system compromise. While the attacker must already possess elevated credentials and network access, successful exploitation can expand their control beyond the application to the host operating system itself.
CVE-2026-55069HIGH 8.7
Kestra is an open-source workflow orchestration platform used to automate complex business processes. A vulnerability in versions prior to 1.3.24 allows attackers with database access to recover administrator credentials through offline password cracking. This is particularly dangerous in Kubernetes environments, where a compromised admin account can be leveraged to steal cluster tokens and secrets, granting access to the entire Kubernetes cluster and its workloads.
CVE-2026-55429HIGH 8.7
Coder, a platform for provisioning remote development environments, contains a critical flaw in how it manages workspace applications and their associated agents. When Coder provisions infrastructure via Terraform, two code paths fail to properly validate ownership of resources: the upsert logic overwrites agent assignments without checking workspace boundaries, and the job completion handler accepts agent IDs without verifying they belong to the workspace being built. An attacker with template author or provisioner operator privileges could exploit this to reassign agents across workspaces, potentially gaining unauthorized access to development environments and sensitive resources. The vulnerability is now fixed in recent patch releases.
CVE-2026-55596HIGH 8.7
Plate is a rich-text editor component library that includes AI capabilities and integrates with shadcn/ui. A cross-site scripting (XSS) vulnerability exists in versions 53.0.0 through 53.1.3 where the media embed renderer fails to properly validate URL protocols when processing saved documents. An attacker can craft a malicious Plate document that claims to embed a known video provider (like YouTube) but secretly points to a javascript: URI. When a victim opens this document, the browser executes the injected JavaScript code directly, bypassing the intended security check. This affects applications that allow users to share or exchange Plate documents.
CVE-2026-56223HIGH 8.7
Capgo versions before 12.128.2 have a critical account takeover flaw in their user provisioning system. An attacker with admin access to an enterprise organization can trick the system into merging victim accounts by forging identity provider assertions. This happens because the system doesn't properly verify that the identity provider making the request is authorized to provision users for the victim's domain. Once merged, the attacker gains complete control of the victim's account, organization, and all associated data.
CVE-2026-57983HIGH 8.7
Microsoft Edge (Chromium-based) contains an authorization flaw that allows attackers to bypass a built-in security feature remotely. The vulnerability requires some specific conditions to exploit but does not require user interaction, making it a significant concern for organizations relying on Edge's security controls. An attacker on the network can circumvent the protection without authentication.
CVE-2026-60104HIGH 8.7
Bitwarden Server contains an authentication bypass vulnerability that allows a low-privileged organization member to impersonate another user and steal their vault encryption key. The flaw stems from insufficient validation when creating a Trusted Device Encryption authentication request—the system fails to confirm that the email address in the request belongs to the person making it. An attacker can craft a request targeting a victim, bind it to their own public key, and once a legitimate user approves it, the attacker can retrieve the victim's vault key and gain full account access from an unauthenticated endpoint. This is particularly dangerous because it requires only basic organizational membership and minimal user interaction.
CVE-2026-6211HIGH 8.7
WEOLL, a platform from Global IT Informatics Services Inc., contains a file upload vulnerability that allows authenticated users to upload malicious files without proper type validation. Because access controls are not correctly enforced on the upload functionality, an attacker with valid credentials can circumvent restrictions and upload dangerous file types, potentially gaining unauthorized access to system features or data they should not be able to reach. The vulnerability affects versions 2.0.9 through 3.2.45.32 and requires an active user session plus user interaction to exploit.
CVE-2026-6552HIGH 8.7
GitLab Enterprise Edition contains an authorization flaw in its Group SAML identity management system that allows a user with Owner role in a group to hijack the GitLab account of another group member. The vulnerability exists across multiple release branches and requires authenticated access at a high privilege level. An attacker exploiting this could gain full control of a victim's account, including access to their repositories, projects, and any resources tied to that account.
CVE-2026-6896HIGH 8.7
GitLab Enterprise Edition contains a cross-site scripting (XSS) vulnerability that allows an authenticated developer to inject malicious scripts into another user's browser session. An attacker with developer-level permissions can craft input that bypasses sanitization controls, causing arbitrary JavaScript to execute when a victim visits a affected page. This requires user interaction—the victim must click a link or visit a page containing the injected payload—but once triggered, the attack can steal session tokens, modify page content, or perform actions on behalf of the victim.
CVE-2026-7313HIGH 8.7
Progress Sitefinity contains a credential exposure vulnerability affecting versions 8.0.5700 through 13.3.7652. An authenticated attacker with backend administrative privileges can retrieve plaintext credentials used by Sitefinity to connect to the Sitefinity Insight analytics service. The vulnerability only manifests when Insight integration is active and non-standard site configuration is in place. While it requires existing backend access and specific preconditions, successful exploitation yields valid service account credentials that could be leveraged for lateral movement or unauthorized data access.
CVE-2026-7574HIGH 8.7
Claude Desktop's Cowork VM image loading mechanism performs a lightweight validation check—confirming the file exists and contains an expected version marker—but stops short of verifying that the image contents remain unmodified. An attacker who gains unprivileged code execution within the macOS user's session can tamper with the VM root filesystem image and cause it to be trusted and booted on the next Cowork VM launch. This allows persistent code execution inside the virtualized environment and potentially compromises any directories mounted from the host.
CVE-2026-9024HIGH 8.7
A stored cross-site scripting (XSS) vulnerability exists in DELMIA Service Process Engineer's Process Experience Studio component across multiple releases. An attacker with user-level access can inject malicious scripts that persist in the application, executing in the browsers of other users who view the affected content. This allows unauthorized access to sensitive data or actions performed on behalf of victims within their authenticated sessions.
CVE-2019-25719HIGH 8.6
Dräger's Infinity patient monitoring systems contain flaws in how they handle network communications that allow an attacker on the same network segment—or within wireless range—to interfere with device operation. An attacker could change critical alarm settings, flood the device with traffic to force a reboot, or disrupt network connectivity. The vulnerability affects Infinity Acute Care System and Standalone M540 monitors running software versions VG4.1.1, VG4.0.3, and earlier.
CVE-2022-4992HIGH 8.6
Dräger patient monitors—specifically the Infinity Acute Care System and Standalone M540 models—contain a flaw in how they handle network messages. An attacker on the network can send malicious or fake data to these devices without needing credentials, which could cause them to reboot, lose connectivity, or have their alarm settings altered. This is especially serious in clinical environments where patient monitoring continuity is critical.
CVE-2024-40646HIGH 8.6
Vertex, a management tool for private tracker users, contains a path traversal vulnerability that allows remote attackers to access files outside their intended directory. An attacker can craft malicious requests to navigate the file system and read sensitive data without authentication. The vulnerability affects all versions prior to a specific patch commit and carries a high severity rating due to its ease of exploitation and significant impact on confidentiality.
CVE-2025-69128HIGH 8.6
CVE-2025-69128 is a path traversal vulnerability in EMV JobCareer versions 7.3 and earlier that allows an attacker to access files and directories outside of their intended scope. An unauthenticated, remotely located threat actor can exploit this without any special permissions or user interaction, potentially disrupting the availability of the affected system. This type of vulnerability is particularly dangerous because it operates over the network and requires no authentication, making it accessible to anyone with network connectivity to the vulnerable application.
CVE-2025-71261HIGH 8.6
A vulnerability in SUSE Harvester before version 1.8.0 allows a network-level attacker to interfere with the TLS handshake process between the SUSE Virtualization and Rancher Manager components. This interference enables attackers to bypass TLS encryption entirely, exposing communications that should be protected. The vulnerability requires the attacker to have network access between these components but does not require valid credentials or user interaction.
CVE-2025-7737HIGH 8.6
A denial-of-service vulnerability exists in the 10G iSCSI interface component of multiple Hitachi Virtual Storage Platform models. An unauthenticated attacker on the network can trigger a condition that exhausts system resources, causing the iSCSI interface to become unavailable and disrupting storage access for dependent systems. No authentication is required, and the vulnerability can be triggered with minimal complexity, making it relatively straightforward to exploit once network access to the iSCSI interface is established.
CVE-2026-10649HIGH 8.6
Pacemaker, a widely-used high-availability cluster management framework, contains an integer overflow flaw in how it decompresses remote messages. An attacker without credentials can send a malicious compressed message to the CIB (Cluster Information Base) remote listener before authentication completes, triggering memory corruption that crashes the service. This is a denial-of-service vulnerability accessible over the network with no special prerequisites.
CVE-2026-11158HIGH 8.6
A vulnerability in Google Chrome's download handling on macOS allows a local attacker to potentially escape Chrome's sandbox protection using a specially crafted AppleScript command. The issue stems from insufficient validation of user-supplied input. An attacker with local access to an affected Mac could exploit this to break out of the browser sandbox and gain elevated system privileges, though user interaction (such as clicking on a malicious download link or AppleScript trigger) is required.
CVE-2026-11527HIGH 8.6
Config::IniFiles is a Perl library that reads configuration files in INI format. Versions before 3.001000 contain a critical flaw in how they open configuration files: they use Perl's two-argument open() function, which interprets special characters in filenames as shell commands or file redirections. If an attacker can control the filename passed to the library (via the -file argument), they can execute arbitrary OS commands or overwrite files on the system. The vulnerability affects any application using Config::IniFiles that processes untrusted configuration file paths.
CVE-2026-11590HIGH 8.6
WP Support Plus Responsive Ticket System, a WordPress plugin for managing support tickets, contains a SQL injection vulnerability that allows attackers without any credentials to query or manipulate the site's database. The flaw stems from improper handling of user-supplied array keys in database queries. An unauthenticated attacker can craft malicious input to extract sensitive information such as user credentials, customer data, or configuration details. This is a serious risk for any WordPress site running the affected plugin versions.
CVE-2026-12053HIGH 8.6
GitLab Enterprise Edition versions 19.1.0 contain a flaw in Duo Workflows that can leak sensitive information already stored in a project repository. An attacker without authentication can exploit this to read committed data that should remain private, bypassing normal access controls. The issue stems from inadequate filtering of workflow output, allowing sensitive content to be exposed. GitLab fixed this in version 19.1.1.
CVE-2026-12057HIGH 8.6
A vulnerability in Foxit AI allows malicious PDF files to execute arbitrary code on a user's computer. When a PDF contains embedded JavaScript, the application's sandbox—designed to isolate and restrict dangerous operations—fails to block certain risky interfaces. This oversight allows remote code to be loaded and executed with the privileges of the logged-in user, potentially giving an attacker full control of the system.
CVE-2026-13849HIGH 8.6
Google Chrome on Windows contains a flaw in its Chromoting component (the remote desktop feature) that fails to properly validate certain user inputs. An attacker with local access to a machine can exploit this by tricking a user into opening a malicious file, potentially breaking out of Chrome's security sandbox and gaining broader system access. The vulnerability affects Chrome versions prior to 150.0.7871.47 on Windows.
CVE-2026-22312HIGH 8.6
A network-accessible web server on affected devices uses a hardcoded authentication token to protect its REST API. Because the token is constant and embedded in the device, an attacker on the network can bypass authentication entirely, then read system settings, reconfigure the device, and trigger administrative commands like system reboot—all without any credentials or user interaction.
CVE-2026-3326HIGH 8.6
A critical flaw in the Xstore WordPress theme before version 9.7.3 allows attackers to execute SQL injection attacks without any authentication. The vulnerability exists in an AJAX handler that fails to properly sanitize user input before incorporating it into database queries. Because the vulnerable endpoint is accessible to anonymous users, this represents a severe risk to any WordPress site using the affected theme version.
CVE-2026-37232HIGH 8.6
OpenAirInterface5G version 2.4.0 contains a vulnerability in how it calculates radio resource block (PRB) utilization metrics for 5G base stations. An attacker on the network can repeatedly request performance monitoring data via the FlexRIC interface, causing the base station software to crash when certain metric calculations divide by zero. The crash takes down the entire 5G cell, disconnecting all users. This requires no authentication and can be triggered remotely.
CVE-2026-40769HIGH 8.6
A vulnerability in Contact Form Extender for Divi (versions 1.0.6 and earlier) allows attackers to delete arbitrary files from affected websites without needing to log in or authenticate. The flaw stems from insufficient access controls on file deletion functionality, meaning anyone with network access to the site can trigger file removal. This poses a significant availability risk, as attackers could disable websites by deleting critical files or data storage.
CVE-2026-40999HIGH 8.6
Spring Web Services has a security flaw in how it handles WS-Addressing protocol features. When a SOAP request includes a ReplyTo or FaultTo address (instructions on where to send responses), the library automatically connects to those addresses without checking whether they're legitimate. An attacker can craft a malicious SOAP request with a ReplyTo pointing to any server they control, causing your application to make an outbound connection to that server. This is a Server-Side Request Forgery (SSRF) risk: the attacker can probe internal networks, exfiltrate data, or trigger unwanted actions on behalf of your application.
CVE-2026-42089HIGH 8.6
Yeoman Environment, a popular Node.js scaffolding tool, contains a critical flaw in how it handles generator installation. When a project configuration file (such as package.json or a Yeoman config) contains references to generator packages, the vulnerable versions will automatically download and install those packages without asking the user first. An attacker who can control project configuration—through a malicious repository, a compromised dependency, or social engineering—can exploit this to install arbitrary npm packages and execute code on a developer's machine during the initial project setup. The flaw affects versions 2.9.0 through 6.0.0, with a fix confirmed in 6.0.0.
CVE-2026-4249HIGH 8.6
A weakness in how WSO2 API infrastructure products handle event throttling allows attackers to send specially crafted data that crashes or disables the API Gateway without authentication. Once triggered, the system stays down until someone manually restarts it, potentially blocking all API traffic for hours or longer.
CVE-2026-44461HIGH 8.6
Zed, a modern code editor, has a vulnerability in how it constructs commands for remote development over SSH or WSL (Windows Subsystem for Linux). When opening a terminal in a remote session, Zed builds a shell command that includes environment variables, but it doesn't properly escape or validate the names of those variables. An attacker who can inject a malicious environment variable name—such as through project-level terminal settings—can embed shell commands within that name. When the remote shell executes the command, it interprets these embedded instructions, allowing arbitrary code execution on the remote machine under the user's privileges. The flaw has been patched in version 0.227.1.
CVE-2026-44463HIGH 8.6
Zed, a modern code editor, contains a security flaw in how it controls which programs can run in the integrated terminal. An attacker can bypass these permission restrictions by sneaking environment variable assignments into commands that are supposed to be allowed. By manipulating variables like PAGER, an attacker can redirect the editor to run malicious code when it attempts to display output. This affects Zed versions before 0.229.0 and is resolved in that release.
CVE-2026-44465HIGH 8.6
Zed, a modern code editor, contains a critical vulnerability that allows attackers to execute arbitrary commands on a user's system. The flaw exists in how Zed handles Git configuration files when opening project folders. Specifically, attackers can craft a malicious .git/config file that abuses Git's core.fsmonitor setting to trigger code execution. When a user opens a folder containing this malicious configuration in untrusted mode, the attacker's commands run with the user's privileges. This is a local attack vector requiring user interaction, but the impact is severe: full system compromise including data theft, modification, and denial of service.
CVE-2026-44466HIGH 8.6
Zed is a popular code editor that includes a built-in terminal tool with a permission system designed to control which commands users can execute. Researchers discovered that this permission system has a critical flaw: attackers can bypass it using bash arithmetic expansion syntax—specifically the $((...)) construct—to nest arbitrary commands inside allowlisted commands like echo. This means a user could trick Zed into running malicious code even though the underlying command appeared safe. The vulnerability affects all versions of Zed prior to 0.229.0 and has been patched in that release.
CVE-2026-44492HIGH 8.6
Axios, a widely-used HTTP client library for JavaScript, contains a bypass vulnerability in its proxy filtering logic. When developers configure Axios to skip the proxy for certain internal IP addresses (via the NO_PROXY setting), an attacker can circumvent that restriction by using an IPv4-mapped IPv6 address format. For example, if 127.0.0.1 is marked as no-proxy, a request to ::ffff:7f00:0:1 still routes through the proxy, potentially exposing internal services. Node.js resolves these IPv6 addresses back to their underlying IPv4 form, so the request reaches the intended internal host while also passing through the proxy infrastructure—defeating the security boundary the no-proxy list was meant to enforce.
CVE-2026-44697HIGH 8.6
Klever-Go versions prior to 1.7.17 contain a remote denial-of-service flaw that allows any network participant to crash validator nodes by sending a single, small network message (under 50 KB). The vulnerable code mishandles decompression of batched data, causing the receiving node to allocate multiple gigabytes of memory unexpectedly, leading to out-of-memory crashes. Because validators are critical to blockchain operation, an attacker can disrupt the entire Klever network's ability to process transactions and reach consensus. The attack requires no authentication and can be executed from any peer on the network.
CVE-2026-45169HIGH 8.6
Idira Privileged Access Manager Self-Hosted Vault contains a flaw in how it validates certain inputs. When specific conditions and configurations align, an attacker can send specially crafted requests over the network that cause the vault service to crash unexpectedly. While the service is down, legitimate users cannot access their privileged credentials, creating a denial-of-service condition. The vulnerability affects multiple older versions of the product, though patches have been released.
CVE-2026-46273HIGH 8.6
A flaw in the Linux kernel's ibmveth driver can cause certain Power system network adapters to freeze when handling segmented network traffic with unusually small packet sizes. The adapter lockup stops all network traffic until manual intervention, creating a denial-of-service condition. The issue occurs only when the adapter's hardware segmentation feature (GSO) encounters packets smaller than 224 bytes that require splitting into multiple segments. The fix implements a software-based workaround that performs segmentation on the system side instead of relying on the hardware, preventing the adapter freeze.
CVE-2026-46491HIGH 8.6
SimpleSAMLphp-casserver versions before 7.0.3 contain a path traversal vulnerability in their file-based ticket storage mechanism. An attacker can manipulate CAS ticket identifiers to read, deserialize, and potentially delete arbitrary files on the server. The vulnerability exists in public-facing CAS validation endpoints that lack proper input sanitization, making it exploitable without authentication.
CVE-2026-46776HIGH 8.6
Oracle Unified Directory, a directory service component in Oracle Fusion Middleware, contains a network-accessible vulnerability that allows attackers without credentials to read, modify, or delete sensitive directory data. An attacker on the network can exploit this flaw via the LDAP protocol to gain unauthorized access to critical configuration and user information, with limited ability to disrupt service. The vulnerability affects two specific versions: 12.2.1.4.0 and 14.1.2.1.0.
CVE-2026-47139HIGH 8.6
vm2 is a popular Node.js sandbox designed to safely execute untrusted code in an isolated environment. A flaw in versions before 3.11.4 allows bypassing network restrictions by exploiting Node.js internal HTTP modules that the sandbox wasn't designed to block. Even when administrators explicitly deny access to standard network libraries like http and https, attackers can abuse hidden internal modules (_http_client and _http_server) to make outbound requests or create listening sockets, defeating the sandbox's security controls.
CVE-2026-47209HIGH 8.6
vm2 is a sandbox for Node.js that isolates untrusted code. A bug in how it handles property assignments on inherited objects allows attackers to bypass its security boundaries and write dangerous properties to the host system. Instead of properly creating properties on child objects, the sandbox writes directly to the host, letting malicious code inject capabilities that could compromise the Node.js runtime itself.
CVE-2026-47368HIGH 8.6
A path traversal vulnerability in UniFi OS allows attackers already present on a network to read sensitive data from affected devices or cloud instances. The attacker doesn't need valid credentials or user interaction—they only need network access to exploit the flaw and potentially extract configuration files, credentials, or other stored information.
CVE-2026-47389HIGH 8.6
Mastodon instances running older Ruby versions are vulnerable to a DNS-based attack that bypasses network security controls. An attacker can craft DNS responses pointing to private network addresses disguised as IPv6 addresses, tricking Mastodon into making HTTP requests to restricted targets like localhost or cloud metadata services. This could expose sensitive configuration data or internal services normally isolated from the internet.
CVE-2026-47825HIGH 8.6
Spring Cloud Gateway, a widely-used API gateway in Spring Boot microservices architectures, has a header-spoofing vulnerability. In certain configurations, the gateway fails to properly validate the origin of X-Forwarded-For and Forwarded HTTP headers—standard headers used to track client IP addresses through proxy chains. An attacker can craft requests with forged headers to make the gateway believe requests originate from trusted sources, potentially bypassing IP-based access controls, WAF rules, or authentication mechanisms that rely on client IP validation. This affects both synchronous (WebMVC) and asynchronous (WebFlux) implementations.
CVE-2026-47835HIGH 8.6
Spring AI's vector store implementations for Elasticsearch, OpenSearch, and GemFire contain a query injection vulnerability. An attacker can inject specially crafted characters into vector store queries to execute arbitrary database operations without authentication. This affects Spring AI versions 1.0.0–1.0.8 and 1.1.0–1.1.7, with fixes available in 1.0.9 and 1.1.8 respectively.
CVE-2026-47906HIGH 8.6
Adobe Dreamweaver Desktop versions 21.7 and earlier contain a vulnerability stemming from reliance on an outdated or flawed third-party component. An attacker can exploit this flaw by distributing a malicious file—typically a project file or document—that, when opened by a user in Dreamweaver, executes arbitrary code with the privileges of the person running the application. The vulnerability requires user interaction (opening the file) to trigger, but once activated, it grants the attacker full control to read, modify, or delete data, or further compromise the system.
CVE-2026-47907HIGH 8.6
Adobe Dreamweaver Desktop versions 21.7 and earlier contain a flaw that allows an attacker to run malicious code on a victim's computer. The vulnerability exists because Dreamweaver does not properly restrict access to certain functions. An attacker must trick a user into opening a specially crafted file—Dreamweaver itself will not automatically trigger the issue. Once exploited, the attacker gains the ability to execute code with the same permissions as the user running Dreamweaver, potentially compromising sensitive projects, credentials stored locally, or the broader system.
CVE-2026-48285HIGH 8.6
Adobe ColdFusion contains a Server-Side Request Forgery (SSRF) vulnerability that allows attackers to bypass security controls and read sensitive data without any user interaction. Versions 2025.9, 2023.20, and earlier are affected. An attacker on the network can exploit this remotely to make the ColdFusion server fetch or interact with internal resources, potentially exposing confidential information or accessing protected systems behind the application.
CVE-2026-48721HIGH 8.6
Warp, an AI-assisted development environment, contains a security flaw that allows attackers to bypass safety restrictions on dangerous commands. The vulnerability exists because Warp's non-interactive CLI agent checks whether commands are 'blocked' before properly handling environment variables at the start of command strings. An attacker who can control what the agent outputs—such as through a compromised repository or malicious input—can craft commands that evade the safety denylist. This could allow execution of commands that normally require user confirmation. The flaw affects versions from October 2025 through early May 2026 and is patched in the May 2026 release.
CVE-2026-49127HIGH 8.6
Music Player Daemon (MPD) versions before 0.24.11 contain a flaw in how they decode certain audio formats that allows an attacker to crash the service or potentially execute code. An attacker can send specially crafted commands pointing to a malicious audio file, causing the decoder to write more data than the buffer can hold. This overwrites adjacent memory with attacker-controlled data, destabilizing the daemon. No authentication is required—any network-accessible MPD instance is at risk.
CVE-2026-49202HIGH 8.6
Acer Connect M6E 5G devices store multimedia session archives (recordings, transcripts, or similar session data) in a way that allows anyone on the internet to access them without logging in. The problem is made worse by overly permissive CORS settings, which let attackers retrieve these files from a victim's browser during a cross-site attack—turning a confidentiality leak into an active theft vector. An attacker doesn't need special tools or credentials; they can pull sensitive multimedia data remotely.
CVE-2026-49269HIGH 8.6
Apple M1 GPUs have a data-leakage flaw where GPU register contents left behind by one app can be read by another app. Specifically, when a sandboxed app finishes using the GPU, the registers aren't fully cleared. A second sandboxed attacker app can then run shader code that reads those leftover register values—including cryptographic secrets or other sensitive data. The vendor confirms this only affects older M1 hardware; newer Apple Silicon already has hardware fixes in place.
CVE-2026-49991HIGH 8.6
RustFS 1.0.0-beta.4 contains a critical flaw in its Snowball auto-extract feature that allows authenticated users to bypass multi-tenant isolation. An attacker with only the permission to upload files to their own bucket can manipulate file paths during extraction to write files into other users' buckets, compromising data confidentiality and integrity across the entire system. The vulnerability exploits a chain of three separate validation failures in path handling and permission checking.
CVE-2026-50085HIGH 8.6
The Aqara Board service at op-test.aqara.com has a critical flaw where it accepts any MQTT command without verifying the sender's identity, then passes these commands directly to the platform's message broker. An attacker on the internet can send malicious commands that affect Aqara devices without needing credentials. While this vulnerability alone allows partial device compromise, it becomes far more dangerous when combined with three related vulnerabilities that enable complete device takeover.
CVE-2026-50131HIGH 8.6
Fedify, a TypeScript library for building ActivityPub-based federated servers, contains an incomplete validation bypass in its Server-Side Request Forgery (SSRF) defense. The library was previously patched to validate outbound requests against a list of non-public IPv4 ranges, but the validation logic fails to block several special-use and reserved address ranges. An attacker could craft requests to these inadequately-blocked ranges to access internal network resources or local services, potentially leading to information disclosure or lateral movement within infrastructure where Fedify applications are deployed.
CVE-2026-53217HIGH 8.6
A memory synchronization bug in the Linux kernel's Marvell PPv2 network driver can cause the CPU to read stale data from network packets on systems with non-coherent DMA (direct memory access). The hardware writes received network data to a specific memory offset, but the kernel's cache synchronization was reading from the wrong location and size, potentially missing the end of incoming frames. This means an attacker sending crafted network traffic could cause the kernel to process corrupted or incomplete packet data, leading to crashes, information disclosure, or potential code execution.
CVE-2026-53755HIGH 8.6
Crawl4AI is a web crawler tool designed to work well with AI language models. In versions before 0.8.9, the Docker-based API server had a critical flaw: it blocked requests trying to crawl internal IP addresses, but this protection didn't apply to proxy servers. An attacker could send a normal-looking crawl request that appears to target a legitimate external website, but configure a proxy to route the traffic through internal networks or cloud metadata services. Since the Docker API requires no authentication by default, anyone with network access could exploit this to discover internal services or steal cloud credentials.
CVE-2026-54403HIGH 8.6
A path traversal vulnerability in Ubiquiti's UniFi OS allows an attacker with network access to bypass authentication on affected UniFi devices and controllers. The flaw does not require user interaction or special privileges to exploit, meaning an attacker positioned on the same network could potentially gain unauthorized access to UniFi OS instances without valid credentials. This is a serious issue because UniFi devices are commonly used to manage enterprise and residential networks, and compromised access could lead to network reconnaissance, configuration changes, or lateral movement.
CVE-2026-54407HIGH 8.6
UniFi Protect Application contains an authentication bypass vulnerability that allows an attacker with network access to circumvent login controls on certain API endpoints. An unauthenticated attacker on the network can reach these endpoints and perform unauthorized actions without valid credentials, potentially affecting the confidentiality, integrity, and availability of protected systems.
CVE-2026-54408HIGH 8.6
CVE-2026-54408 is a network-based authentication bypass flaw in UniFi Protect that allows an attacker with network access to stream video data without providing valid credentials. The vulnerability stems from improper access control logic and carries a CVSS score of 8.6 (HIGH severity). While an attacker cannot modify system configuration or cause denial of service at scale, the ability to intercept sensitive surveillance footage represents a significant confidentiality breach for organizations relying on UniFi Protect for physical security monitoring.
CVE-2026-54410HIGH 8.6
nanoMODBUS versions up to and including v1.23.0 contain a critical flaw in how the Modbus/TCP server processes incoming network messages. An attacker can send a specially crafted message that causes the application to write data one byte beyond the intended buffer boundary. On systems without memory protection—common in embedded and real-time environments—this overflow can crash the service, leak sensitive data, or allow unauthorized writes to device registers. No authentication is required; an attacker on the network can exploit this remotely.
CVE-2026-54762HIGH 8.6
Traefik, a widely-used reverse proxy and load balancer, has a flaw in how it handles Kubernetes authentication configuration. When an operator sets up BasicAuth or DigestAuth protection on a route via Kubernetes annotations but the system cannot find or read the required authentication credential (the Secret), Traefik logs the error and then publishes the route anyway—without the authentication protection. This means unauthenticated users can access backends that operators believed were protected. The flaw affects versions 3.7.0-ea.1 through 3.7.4 and is corrected in 3.7.5.
CVE-2026-55117HIGH 8.6
A network-accessible vulnerability in UniFi Access Application allows an unauthenticated attacker to read arbitrary files from the host system through path traversal. An attacker positioned on the network can craft requests that bypass directory restrictions, potentially exposing sensitive configuration files, credentials, or application data without requiring any user interaction or prior access privileges.
CVE-2026-55418HIGH 8.6
FastGPT, an open-source AI knowledge base platform, contains a critical authorization flaw that allows attackers to read files belonging to other teams. Two file handlers in versions before v4.15.0-beta5 fail to validate that S3 object keys belong to the requesting user's team. An attacker can bypass these checks by supplying another team's key through the chat-file presign or dataset preview endpoints, gaining unauthorized access to sensitive data. The vulnerability does not require authentication and affects all instances without the fix.
CVE-2026-55441HIGH 8.6
mise, a tool for managing multiple development dependencies (Node, Python, CMake, Terraform), has a critical flaw in versions before 2026.6.4 where tasks can execute arbitrary commands without user consent. The vulnerability bypasses mise's trust feature by loading task files from directories like mise-tasks/ or .mise/tasks/ without ever checking whether the user has approved that directory. When you simply list available tasks or use tab completion—even read-only operations—malicious code embedded in task definitions runs immediately. An attacker needs only to trick a developer into cloning a repository and checking what tasks are available; no additional user action is required beyond that normal workflow step.
CVE-2026-55602HIGH 8.6
http-proxy-middleware, a widely-used Node.js library for routing HTTP requests to backend servers, contains a routing bypass flaw affecting versions 0.16.0 through 2.0.9, 3.0.5, and 4.0.x. When administrators configure the library to route requests based on a combination of hostname and path (host+path rules), the library incorrectly matches incoming requests using substring matching instead of exact matching. An attacker can craft a malicious Host header that contains the configured hostname as a substring—but with extra characters appended—and bypass the intended routing restrictions, directing requests to an unintended backend service. This could expose sensitive data, cause unauthorized access, or trigger unintended application behavior depending on what backends are exposed.
CVE-2026-55604HIGH 8.6
DeepSeek MCP Server versions 1.4.2 through 1.6.x contain a session management flaw that allows attackers to hijack other users' conversation sessions. The vulnerability stems from the server accepting session identifiers without validating that they belong to the requester. An attacker can discover active session IDs through enumeration, then impersonate a victim by reusing that session ID in subsequent requests, gaining access to the victim's chat history and context. This is a session fixation and privilege escalation issue that requires no authentication or user interaction to exploit.
CVE-2026-55638HIGH 8.6
9Router, an AI-based router designed to optimize and reduce token consumption for language model API calls, contains an authentication bypass vulnerability in versions before 0.5.2. The vulnerability exists because the application's URL protection rules in its dashboard guard don't cover the /codex endpoint. An attacker can exploit this gap by sending unauthenticated requests directly to /codex paths, which are later rewritten internally to /api/v1/responses. This rewriting happens after the authentication checks, allowing the attacker to bypass API-key verification and trick the server into making requests to upstream LLM providers using the operator's stored credentials. The impact ranges from data exposure to potential service disruption.
CVE-2026-56261HIGH 8.6
Crawl4AI, a web crawling and data extraction framework, contains a vulnerability that allows attackers to trick the application into making requests to internal services it shouldn't access. When you submit a crawling or LLM job, you can specify a webhook URL where results should be sent back. The vulnerability exists because the application doesn't properly validate these URLs—an attacker can provide a webhook address pointing to private networks, internal IP ranges, or cloud metadata services. This causes the Crawl4AI server to reach out to those internal systems on behalf of the attacker, potentially leaking sensitive information like cloud credentials or internal service data.
CVE-2026-56266HIGH 8.6
Crawl4AI, a web scraping and content processing tool, contains a server-side request forgery (SSRF) vulnerability that allows unauthenticated attackers to make the server fetch arbitrary URLs. By crafting IPv6-mapped IPv4 addresses, attackers can bypass the application's blocklist protections and reach internal services or cloud metadata endpoints—such as AWS Instance Metadata Service—that should be inaccessible. This affects versions prior to 0.8.7.
CVE-2026-56285HIGH 8.6
Nitter, a lightweight Twitter/X interface, contains a critical flaw in its video media proxy feature. The vulnerability stems from two fundamental security oversights: the service fails to restrict which URLs its proxy can fetch, and it relies on a hardcoded cryptographic key that's the same across all installations. This combination allows attackers to trick Nitter into retrieving content from anywhere on the internet—including sensitive cloud metadata services and internal corporate systems—without needing any authentication. An attacker merely needs to compute a valid signature using the known default key, making the attack trivial to execute.
CVE-2026-57573HIGH 8.6
Crawl4AI, an open-source web crawler designed for large language models, contains a server-side request forgery (SSRF) vulnerability in versions before 0.9.0. When deployed as a Docker API server, the application validates destination URLs on its standard crawl endpoint but fails to perform the same check on its streaming variant. This allows unauthenticated attackers to submit requests to the streaming path that direct the server to fetch and stream back content from internal networks, private IP ranges, or link-local addresses that should be inaccessible to external clients.
CVE-2026-57877HIGH 8.6
GeoVision's network video recording devices (GV-LPC2011 and GV-LPC2211, version 1.12 and earlier) contain a format string vulnerability in their login service that can be exploited without authentication. An attacker can send specially crafted login requests over the network to trigger memory corruption, leak sensitive information, or crash the service. The flaw stems from unsafe handling of user-supplied data when constructing log messages during login attempts.
CVE-2026-58049HIGH 8.6
FFmpeg's RASC video decoder contains a flaw that allows memory corruption when processing specially crafted video files. The decoder reads and writes data at incorrect memory locations before properly checking boundaries, and miscalculates region sizes in a way that permits access beyond allocated buffer space. An attacker can craft a malicious video file that, when decoded, writes to and reads from memory outside intended bounds, potentially enabling code execution or system compromise.
CVE-2026-58192HIGH 8.6
Appium, a popular cross-platform automation framework used for testing mobile and web applications, contains a critical path traversal vulnerability in its storage plugin. An unauthenticated attacker can send specially crafted requests to bypass the storage directory boundary using directory traversal sequences (../) and delete arbitrary files or directories on the affected system. The vulnerability exists in versions prior to 1.1.6 and has been patched in the latest release.
CVE-2026-59707HIGH 8.6
LocalAI, an open-source AI framework, contains a vulnerability that allows unauthenticated attackers to make the server fetch data from internal or private network addresses. By sending a specially crafted request to a specific endpoint, an attacker can force the LocalAI server to reach out to addresses it normally shouldn't access—such as internal services or loopback addresses—and extract sensitive information from error messages. This type of attack, known as server-side request forgery (SSRF), is particularly dangerous in cloud or containerized environments where internal services may be exposed on predictable addresses.
CVE-2026-60105HIGH 8.6
Monsta FTP versions before 2.14.5 contain a server-side request forgery (SSRF) vulnerability that allows unauthenticated attackers to make the server issue HTTP requests to internal services on behalf of the attacker. The flaw stems from an incomplete IP blocklist check that fails to recognize IPv4 addresses when they are embedded within IPv4-mapped IPv6 address formats (e.g., ::ffff:192.168.1.1). An attacker can exploit this by obtaining a CSRF token from a public endpoint, then crafting a fetchRemoteFile request that causes the server to retrieve data from internal services—such as cloud metadata endpoints—and write the responses to an FTP location under the attacker's control, potentially exposing sensitive credentials.
CVE-2026-7862HIGH 8.6
A flaw in the Eupago Gateway For WooCommerce plugin (versions before 4.7.2) allows anyone on the internet to process refunds on any order without logging in. Attackers can exploit this to steal money by redirecting refunds to accounts they control. The vulnerability stems from missing access controls on the refund handler.
CVE-2025-53828HIGH 8.5
A vulnerability in SharePoint for ownCloud allows administrators with access to the system to execute arbitrary code by exploiting a server-side request forgery (SSRF) flaw. This is a privilege escalation risk—an admin account in the hands of an attacker, or a malicious insider with admin rights, can use the SharePoint app to run commands on the ownCloud server. Organizations running ownCloud 10 below version 10.15.3 are affected and should upgrade immediately to close this attack vector.
CVE-2025-69135HIGH 8.5
A SQL injection vulnerability exists in the WordPress Events Calendar plugin version 2.7.2 and earlier. An authenticated user with subscriber-level permissions can inject malicious SQL commands through the Events Schedule feature, potentially exposing sensitive database information. This vulnerability requires an attacker to have valid WordPress credentials but does not require user interaction, making it a significant risk for multi-user WordPress installations.
CVE-2026-10055HIGH 8.5
Eclipse Theia versions 1.26.0 and later contain a server-side request forgery (SSRF) vulnerability in the `/services/request-service` RPC endpoint. An attacker who gains access to a Theia service connection can trick the server into making HTTP requests to internal systems and returning the responses. This allows circumvention of network boundaries to access resources like localhost services, cloud metadata endpoints, and administrative interfaces that should not be accessible from the internet. The risk is highest in multi-tenant or publicly-exposed Theia deployments.
CVE-2026-10129HIGH 8.5
IBM Langflow OSS versions 1.0.0 through 1.9.3 contain a Server-Side Request Forgery (SSRF) protection bypass flaw. Attackers with basic flow-author privileges can exploit the API Request component by enabling redirect following on a benign URL that chains to internal or localhost services. The application validates the first URL but ignores redirect destinations, allowing unauthorized access to internal systems, metadata endpoints, and private network resources. This can expose credentials, API tokens, and sensitive administrative data.
CVE-2026-11594HIGH 8.5
IBM WebSphere Application Server versions 8.5 and 9.0 contain a cross-site scripting (XSS) flaw in their administrative console. An attacker can inject malicious scripts that execute in the context of an administrator's browser session, potentially allowing unauthorized actions on the application server. The vulnerability requires network access to the console and user interaction, but once triggered, could compromise administrative functions and server integrity.
CVE-2026-11714HIGH 8.5
IBM WebSphere Application Server Liberty contains a server-side request forgery (SSRF) vulnerability when the apiDiscovery feature is enabled. This flaw allows an authenticated attacker to make the affected server issue requests to arbitrary internal or external systems on behalf of the attacker, potentially accessing sensitive resources or data that should not be directly reachable. The vulnerability requires valid user credentials to exploit but can affect the confidentiality of data and the integrity of internal systems.
CVE-2026-12975HIGH 8.5
Apicurio Registry contains an insecure XML parser configuration that allows attackers to upload malicious XML documents. When processed, these documents can trigger outbound requests to attacker-controlled servers (SSRF), extract sensitive data from internal systems, or exhaust server resources through entity expansion attacks. The vulnerability requires write access to upload artifacts, though default configurations may not enforce authentication.