CVEs of 2026.
Every CVE published in 2026 that we hold, ranked by exploit odds.
-
CVE-2026-108417
Out-of-bounds read in Remote Desktop Client allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-116922
Access of resource using incompatible type ('type confusion') in Microsoft Office PowerPoint allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-108419
Out-of-bounds read in Remote Desktop Client allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-96988
Out-of-bounds read in Windows DNS allows an authorized attacker to disclose information over a network.
-
CVE-2026-96970
Out-of-bounds read in Windows DNS allows an authorized attacker to disclose information over a network.
-
CVE-2026-113386
Roundcube Webmail before 1.6.20 and 1.7.x before 1.7.5 are affected by IMAP command injection. A remote authenticated attacker can exploit this via CRLF sequences injected into the _uid parameter of the mail/headers and mail/viewsource actions to execute arbitrary IMAP commands.
-
CVE-2026-118146
Incorrect authorization in Active Directory Domain Services allows an authorized attacker to execute code over a network.
-
CVE-2026-95628
Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability in Apache Camel Docling component. This issue affects Apache Camel: from 4.15.0 before 4.18.5, from 4.19.0 before 4.22.2. DoclingProducer builds the command line of the external docling tool and appends to it the custom arguments supplied in the CamelDoclingCustomArguments header, a list of strings whose flags the fix for CVE-2026-40047 restricted to an allowlist. That fix also introduced validatePathSafety to keep path-like argument values inside the intended directory, but the method only looked for relative traversal: it rejected a value containing a traversal sequence such as ../ and a value whose normalized form still held a .. element. It never checked whether a value was an absolute path, and an absolute path without traversal segments contains neither, so it passed unchanged. Every path-like value went through this check, whether it was a separate list element following its flag or was attached to the flag after an equals sign, so an absolute path was accepted in either form, for example as the directory given to the allowlisted --artifacts-path flag, which tells docling where to find its model artifacts. Such a value reached the docling command line as given, so a party able to choose the custom arguments could point the docling process at any directory the Camel process can access, outside the working directory that the check was meant to keep path values in, which is the outcome the CVE-2026-40047 fix set out to prevent. Only the CLI mode, which is the default, passes custom arguments to docling. The header belongs to the Camel header namespace, which consumers applying Camel's default header filtering, such as the JMS and Kafka consumers, remove from incoming messages, so the issue is reached where a route forwards externally influenced values into the header or where a consumer maps incoming headers without that filtering. Users are recommended to upgrade to version 4.23.0, which fixes the issue. If users are on the 4.22.x releases stream, then they are suggested to upgrade to 4.22.2. If users are on the 4.18.x releases stream, then they are suggested to upgrade to 4.18.5. The 4.15.x to 4.17.x and 4.19.x to 4.21.x release lines reached end of life before this issue was resolved and will not receive a fix, so deployments on them must move to a supported release. For deployments that cannot upgrade immediately, do not build the CamelDoclingCustomArguments header from values that originate outside the deployment, remove that header from messages that arrive from outside before they reach a docling endpoint, and make sure that consumers which map incoming message headers drop the Camel header namespace, so that only the route itself chooses the custom arguments.
-
CVE-2026-95629
Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability in Apache Camel Docling component. This issue affects Apache Camel: from 4.15.0 before 4.18.5, from 4.19.0 before 4.22.2. DoclingProducer runs the external docling tool in CLI mode, the default, and tells it where to write the converted document through the --output flag. When the CamelDoclingOutputFilePath header was set, addOutputDirectoryArguments handed its value to that flag verbatim: the value was not normalized, an absolute path or a path containing .. segments was accepted as given, and no endpoint option could confine it to a directory. The custom-argument validation added by the fix for CVE-2026-40047 refuses the --output and -o flags in the CamelDoclingCustomArguments header and tells callers to choose the output directory with this header instead, so the one value that decides where docling writes reached the command line without any of the checks applied to custom arguments. Docling writes the conversion result, as Markdown, HTML, JSON or text, into that directory, naming the file after the input file, and the producer afterwards removes only its own temporary directory, so the file is left in place. A party able to set the header could therefore have conversion output created in any directory the Camel process can write to. The content of such a file is the conversion of the document carried by the message, so it is shaped by whoever supplies that document without being written byte for byte, and when the document content arrives in the message body the producer stores it under a random temporary name, which the output file then carries. The header belongs to the Camel header namespace, which consumers applying Camel's default header filtering, such as the JMS and Kafka consumers, remove from incoming messages, so the issue is reached where a route forwards externally influenced values into the header or where a consumer maps incoming headers without that filtering. The docling-serve mode does not use the header. Users are recommended to upgrade to version 4.23.0, which fixes the issue. If users are on the 4.22.x releases stream, then they are suggested to upgrade to 4.22.2. If users are on the 4.18.x releases stream, then they are suggested to upgrade to 4.18.5. The fix adds an outputBaseDirectory endpoint option that is unset by default, so after upgrading, set it on every docling endpoint whose CamelDoclingOutputFilePath header can carry a value influenced from outside the deployment; without it the header is normalized but can still name any directory. The 4.15.x to 4.17.x and 4.19.x to 4.21.x release lines reached end of life before this issue was resolved and will not receive a fix, so deployments on them must move to a supported release. For deployments that cannot upgrade immediately, do not build the CamelDoclingOutputFilePath header from values that originate outside the deployment, remove that header from messages that arrive from outside before they reach a docling endpoint, and make sure that consumers which map incoming message headers drop the Camel header namespace, so that only the route itself chooses the output directory.
-
CVE-2026-118728
Out-of-bounds read in Microsoft Office Excel allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-118300
Out-of-bounds read in Microsoft Office Excel allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-118157
Weak authentication in Microsoft Exchange Server allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-112320
The GSS-RPC transport layer in MIT Kerberos 5 (aka krb5) 1.4 and later does not release the previous authentication context when a client switches from an RPCSEC_GSS or AUTH_GSSAPI flavor to a non-GSS flavor on the same TCP connection. This allows remote attackers to cause a denial of service (memory exhaustion in kadmind) by repeatedly establishing and abandoning GSS authentication contexts.
-
CVE-2026-96078
Heap-based buffer overflow in Microsoft Color Management allows an unauthorized attacker to execute code over a network.
-
CVE-2026-95901
Double free in Windows Microsoft DirectMusic allows an unauthorized attacker to execute code over a network.
-
CVE-2026-78674
The BPMN System REST APIs do not sufficiently validate input parameters, allowing authenticated users to submit crafted requests that can lead to the execution of arbitrary commands on the server. An authenticated user can leverage this vulnerability to execute arbitrary commands with the privileges of the product's runtime account. The scope of potential damage is contingent upon the permissions assigned to this account.
-
CVE-2026-121663
Incomplete list of disallowed inputs in Visual Studio Code allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-119969
Heap-based buffer overflow in Graphic Fonts allows an unauthorized attacker to execute code over a network.
-
CVE-2026-44266
Elo is a purely functional expression language for no-code tools. Prior to 1.0.1, the Ruby backend in src/compilers/ruby.ts embeds attacker-controlled text in Ruby double-quoted strings without escaping #, so Ruby interpolation forms in the string-literal emitter or the labelled subtype-constraint branch become executable Ruby. The affected entry points include the /try playground with the Ruby target, bin/eloc -t ruby -x, and defaultFormats.elo.parse() followed by Ruby emission. A failing labelled subtype constraint evaluates its message, while an ordinary string literal is evaluated when the compiled Ruby runs, allowing Ruby code and shell command execution through backticks. JavaScript, Python, and SQL emission are not affected by this Ruby-specific path. This issue is fixed in version 1.0.1.
-
CVE-2026-87862
Importing a `includeSubDomains` parent HSTS entry before a longer-lived child entry, silently drops the child. Consequently, when the parent expires, the child's valid HSTS policy is lost, causing curl to issue unencrypted HTTP requests rather than upgrading to HTTPS.
-
CVE-2026-112808
Exposure of sensitive system information to an unauthorized control sphere in Windows Print Spooler Components allows an authorized attacker to disclose information over a network.
-
CVE-2026-118191
The Photo Gallery, Sliders, Proofing and Themes – NextGEN Gallery plugin for WordPress is vulnerable to arbitrary file deletion due to insufficient file path validation in the handle_job function in all versions up to, and including, 4.5.5 This makes it possible for authenticated attackers, with custom-level access and above, to delete arbitrary files on the server, which can easily lead to remote code execution when the right file is deleted (such as wp-config.php). The 'NextGEN Manage gallery' capability required to reach the vulnerable endpoint is granted to Administrators by default but can be delegated to lower-privileged roles via the plugin's Roles screen.
-
CVE-2026-112690
This issue was addressed with improved input validation. This issue is fixed in iOS 26.7.2 and iPadOS 26.7.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, macOS Sequoia 15.8.2, macOS Tahoe 26.7.2, visionOS 27.2. A remote attacker may be able to break out of the WebContent sandbox.
-
CVE-2026-116697
Null pointer dereference in Reliable Multicast Transport Driver (RMCAST) allows an unauthorized attacker to deny service over a network.
-
CVE-2026-115893
Wasmtime is a runtime for WebAssembly. Prior to 36.0.19, 48.0.7, and 49.0.4, the Cranelift compiler backend on aarch64 can compile a Wasm br_table instruction with more than approximately 256,000 targets so that it has an incorrect conditional branch target. The branch can then jump to arbitrary other code at a user-controlled offset, allowing a user-controlled Wasm module to escape the sandbox on aarch64 engines using the default Cranelift backend. This issue is fixed in versions 36.0.19, 48.0.7, and 49.0.4.
-
CVE-2026-114169
Heap-based buffer overflow in Windows Raw Image Extension allows an unauthorized attacker to execute code over a network.
-
CVE-2026-124218
The Drag and Drop File Upload for Elementor Forms plugin for WordPress is vulnerable to arbitrary file deletion due to insufficient file path validation in the remove_files function in all versions up to, and including, 1.6.2 This makes it possible for unauthenticated attackers to delete arbitrary files on the server, which can easily lead to remote code execution when the right file is deleted (such as wp-config.php). This vulnerability is reachable when the "yeepdf_not_save" privacy option is enabled, a commonly used production setting that triggers the remove_files() cleanup path.
-
CVE-2026-111752
Angular is a development platform for building mobile and desktop web applications using TypeScript/JavaScript and other languages. Prior to 20.3.33, 21.2.25, and 22.2.1, @angular/router in Node.js server-side rendering fails to require auxiliary outlet URL segments to match routes configured for their outlet names. Crafted requests containing multiple unconfigured empty-path outlet groups can repeatedly evaluate the route configuration, construct duplicate ActivatedRouteSnapshot trees, execute canMatch guards and resolvers, and copy route parameters and queryParams for each outlet. An unauthenticated remote attacker can exhaust the Node.js heap or CPU and terminate an SSR worker when SSR is enabled, a route level contains an empty-path route or named outlet, upstream infrastructure forwards the request target, and the worker has a constrained heap. The allocation is transient rather than a persistent leak, and client-side SPAs and build-time prerendering are not affected. This issue is fixed in versions 20.3.33, 21.2.25, and 22.2.1.
-
CVE-2026-123621
Usermin through 2.570 contains a path traversal vulnerability in mailbox/detachall.cgi that allows remote unauthenticated attackers to write files outside the temporary directory via unsanitized attachment filenames. Attackers can email attachments with ../ sequences in filenames, and when the recipient downloads all attachments, files like ~/.forward are overwritten, enabling command execution.
-
CVE-2024-58408
Microsoft jxrlib through 1.1 and the mircomir jxrlib fork before 1.3.1 contain a global buffer over-read vulnerability in initLookupTables() in image/decode/strdec.c caused by an unchecked bit depth index. Attackers can supply a crafted JPEG XR file with an out-of-range bit depth in the image header to read beyond the cbChannels array, causing crashes or memory disclosure.
-
CVE-2024-58407
Microsoft jxrlib through 1.1 and the mircomir jxrlib fork before 1.3.1 contain a global buffer over-read vulnerability in DecodeSignificantRun() in image/decode/segdec.c due to an unchecked run length. Attackers can supply a crafted JPEG XR file with an out-of-range iMaxRun value that indexes gSignificantRunBin[] out of bounds, causing memory disclosure or crashes during decoding.
-
CVE-2026-116721
Use of uninitialized resource in Microsoft Office Word allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-108424
Out-of-bounds read in Windows GDI allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-120003
Heap-based buffer overflow in Microsoft Office allows an unauthorized attacker to execute code over a network.
-
CVE-2026-121933
A flaw has been found in Linksys E5350 1.0.00.037. This impacts the function save_settings of the file /wizard/apply.cgi of the component Setup Wizard. Executing a manipulation of the argument wizard_router_pwd can lead to missing authentication. The attack may be launched remotely. The exploit has been published and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
-
CVE-2026-111754
Angular is a development platform for building mobile and desktop web applications using TypeScript/JavaScript and other languages. Prior to 20.3.33, 21.2.25, and 22.2.1, ServerPlatformLocation in Angular server-side rendering checks for protocol-relative URLs before WHATWG URL normalization. A request path containing a popping dot segment followed by multiple slashes can bypass the raw resolveUrl check and normalize to a pathname beginning with two slashes. When Angular SSR detects a mismatch with the router's final location, it can place that pathname in an HTTP Location header, causing a browser to navigate to an attacker-controlled origin. Affected applications use @angular/platform-server and @angular/ssr and either expose compatible parameter or child route shapes or pass user-controlled navigation targets through infrastructure that preserves the relevant dot segments or semicolons. Pure client-side SPAs are not affected. This issue is fixed in versions 20.3.33, 21.2.25, and 22.2.1.
-
CVE-2026-96972
Out-of-bounds read in Windows DNS allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-119970
Out-of-bounds read in Windows DNS allows an authorized attacker to disclose information over a network.
-
CVE-2026-116943
Buffer over-read in SQL Server allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-119971
Out-of-bounds read in Windows DNS allows an authorized attacker to disclose information over a network.
-
CVE-2026-83684
Buffer over-read in Windows iSCSI Target Service allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-95882
Stack-based buffer overflow in Windows Microsoft DirectMusic allows an unauthorized attacker to execute code over a network.
-
CVE-2026-67544
Heym builds and runs agentic systems and business-process automations on self-hosted infrastructure. Prior to version 0.0.71, Heym's workflow engine accepts an arbitrary Python code string in the playwrightCode field of a Playwright node. This string is written to a temporary .py file and executed directly by the backend process via subprocess.Popen([sys.executable, script_path]) with no sandboxing, AST validation, or role gate. Any authenticated user — including a freshly registered account — can achieve full remote code execution (RCE) on the backend host, running as the service process user (observed as root in the default Docker deployment). This issue has been patched in version 0.0.71.
-
CVE-2026-118297
Heap-based buffer overflow in Microsoft Office Excel allows an unauthorized attacker to execute code over a network.
-
CVE-2026-79878
A command injection vulnerability in the AOS-8 command line interface accessed by the PAPI protocol could lead to unauthenticated remote code execution by an adjacent attacker sending specially crafted PAPI packets destined to the PAPI UDP port 8211. Successful exploitation could allow an attacker to execute arbitrary code as a privileged user on the underlying operating system.
-
CVE-2026-121680
Exposure of sensitive information to an unauthorized actor in Windows NTLM allows an unauthorized attacker to perform spoofing over a network.
-
CVE-2026-112757
Exposure of sensitive information to an unauthorized actor in Windows Secure Kernel Mode allows an authorized attacker to disclose information locally.
-
CVE-2026-67646
Heap-based buffer overflow in SQL Server allows an authorized attacker to execute code over a network.
-
CVE-2026-115414
OS command injection in the Terraform hook in AWS SAM CLI before 1.166.2 might allow an actor who can influence the Terraform configuration or plan file consumed by a build to execute arbitrary commands as the build user on the host running the build via a crafted Terraform resource address. To remediate this issue, users should upgrade to version 1.166.2 or later.
-
CVE-2026-125350
Use after free in XML in Google Chrome prior to 157.0.8103.50 allowed a remote attacker to execute arbitrary code inside the sandbox via a crafted HTML page. (Chromium security severity: High)
CVE records published in 2026 (ids starting CVE-2026-), sorted by how likely they are to be exploited.