The internet is broken right now.
68 critical CVEs without a fix in the last 48 hours. Developers are crying quietly into their IDEs. 151 more already have a fix that someone still has to apply.
-
CVE-2026-118208
Improper access control in Microsoft Exchange Server allows an authorized attacker to perform spoofing over a network.
-
CVE-2026-114171
User interface (ui) misrepresentation of critical information in Visual Studio Code allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-105560
Release of invalid pointer or reference in Windows DHCP Server allows an authorized attacker to execute code over a network.
-
CVE-2026-116934
Untrusted pointer dereference in SQL Server allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-113897
This issue was addressed by removing the vulnerable code. This issue is fixed in Safari 27.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, visionOS 27.2. A remote attacker may be able to break out of Web Content sandbox.
-
CVE-2026-128027
FrankenPHP Docker images before 1.1.11 were rebuilt only when new base image tags were created, so security fixes that Alpine, PHP, and Go applied under existing tags were not propagated. Remote attackers could exploit known flaws in the stale system libraries, such as CVE-2025-15467 in libcrypto3, to execute code.
-
CVE-2026-124020
Control Web Panel (CWP) before webpanel_version 1.30.6 contains a host header injection vulnerability that allows unauthenticated attackers to manipulate the password-recovery reset URL by injecting the HTTP Host header into the /v1/lostpass/ route's token generation process. Attackers can lure an automated mail-security scanner into visiting the poisoned reset link to exfiltrate the live recovery token, replay it at /login/index.php?acc=newpass to set a new password, and subsequently abuse the File Manager and Crontab modules to achieve command execution as the targeted hosting account.
-
CVE-2026-86410
A flaw in the curl command line tool's logic makes it only set SSH host verification options when interacting directly using SFTP or SCP, and not as a result of following a redirect.
-
CVE-2026-118741
Improper authentication in ASP.NET Core allows an unauthorized attacker to elevate privileges over a network.
-
CVE-2026-95663
Insufficient Session Expiration vulnerability in Apache Camel Keycloak component. This issue affects Apache Camel: from 4.16.0 before 4.18.5, from 4.19.0 before 4.22.2. When token introspection is enabled on the camel-keycloak KeycloakSecurityPolicy (useTokenIntrospection=true), the security processor validates the access token of a request by sending it to Keycloak's OAuth 2.0 token introspection endpoint, and accepts the token while the response reports it as active. The introspection endpoint reports a token as inactive once its exp (expiration) time has passed, so on this path the active flag is what enforces token expiry. Introspection results are cached by default (introspectionCacheEnabled=true, introspectionCacheTtl=60 seconds), and both cache implementations shipped with the component, ConcurrentMapTokenCache and CaffeineTokenCache, expired an entry a fixed time after it was inserted, without bounding that lifetime by the exp claim that the cached response already carried. KeycloakSecurityProcessor in turn checked only the active flag of the result, on every path that uses introspection, and never compared exp with the current time. A response cached shortly before the token expired was therefore served to later requests carrying the same token for up to the configured time-to-live after the token's expiry, and those requests were admitted with the roles and permissions the expired token had carried, although RFC 7662, which defines token introspection, states that a response containing exp must not be cached beyond that time. The access token is supplied by the caller with each request, typically as a bearer token in the Authorization header, so a caller holding a token that was valid moments earlier could keep using it after it expired, for as long as the cached entry lived. The window is bounded by introspectionCacheTtl, which is 60 seconds by default and 300 to 600 seconds where the documentation's recommendation for internal services was followed, and it opens only for a token that the same Camel process introspected and cached while the token was still active. The issue grants nothing beyond what the token already carried: it allows no token forgery and no privilege escalation, and token introspection is not enabled by default, which is why the severity is assessed medium. The local JWT verification path, used when introspection is disabled, is not affected by this issue. 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 4.16.x, 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, set introspectionCacheEnabled=false on the KeycloakSecurityPolicy, so that every request is checked against the introspection endpoint, which reports an expired token as inactive, at the cost of one introspection call per request. Lowering introspectionCacheTtl only narrows the window and does not close it.
-
CVE-2026-107155
A permissive cross-domain policy vulnerability has been identified in the SonicOS management interface, which in specific conditions could potentially enable a remote unauthenticated attacker to perform unauthorized operations.
-
CVE-2026-105426
Improper input validation in Ivanti EPMM before version 12.11.0.0, 12.10.0.1, and 12.9.0.3 allows a remote authenticated attacker with admin privileges to set arbitrary Linux kernel parameters as root.
-
CVE-2026-96196
Incomplete list of disallowed inputs in Microsoft Edge (Chromium-based) allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-112408
Missing verification of source of a communication channel in the default_bucket() function in Amazon SageMaker Python SDK V3 before 3.10.0 and V2 before 2.257.3 might allow remote actors to achieve code execution in the target SageMaker execution environment and obtain IAM credentials via pre-creation of a predictably named S3 bucket that the SDK uses without ownership validation. To remediate this issue, users should upgrade to Amazon SageMaker Python SDK 3.10.0 (V3) or 2.257.3 (V2).
-
CVE-2026-86413
When using an *ambient user* in a first transfer doing Negotiate authentication and asked to follow redirects, curl could pass on its credential details to an unrelated second origin when asked to redirect there.
-
CVE-2026-119976
External control of file name or path in Visual Studio Code - GitHub Pull Requests and Issues Extension allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-112944
launch-editor allows users to open files with line numbers in editor from Node.js. Prior to 2.14.2, when an attacker can create a filesystem entry beginning with plus or minus in the working directory, an affected editor is selected, and the path can remain relative or is converted to a relative WSL path, launchEditor forwards the path without forcing it to be interpreted as a positional filename. Editor option parsing then treats the attacker-controlled filename as a command-line flag, which can cause unintended file overwrite or arbitrary command execution by the selected editor. The affected editors are Vim and gVim, MacVim, Emacs and Emacsclient, and Notepad++ when a plugin exposes dangerous operations; WSL path conversion keeps Vite and newer middleware callers exposed. This issue is fixed in version 2.14.2.
-
CVE-2026-121660
Incomplete list of disallowed inputs in Visual Studio Code allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-116929
Execution with unnecessary privileges in SQL Server allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-119991
Weak authentication in Microsoft Configuration Manager allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-112682
An input validation 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 Tahoe 26.7.2, visionOS 27.2, watchOS 27.2. A remote attacker may be able to cause unexpected app termination.
-
CVE-2026-118739
Missing release of memory after effective lifetime in Windows DHCP Client allows an unauthorized attacker to deny service over an adjacent network.
-
CVE-2026-17410
IBM API Connect 10.0.8.0 through 10.0.8.12 and 12.1.0.0 through 12.1.1.2 could allow a remote authenticated attacker to access and modify other tenants' portal data due to improper neutralization of argument delimiters.
-
CVE-2026-107929
When `CURLOPT_INTERFACE` (or the command line tool's `--interface` option) is set with one of the documented prefixes (`if!`, `host!`, `ifhost!`), that binding is not part of the connection reuse key, so a later transfer asking for a different source can silently reuse an existing connection and go out from the earlier transfer's source address
-
CVE-2026-119998
Product ui does not warn user of unsafe actions in Microsoft Office allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-122177
Improper link resolution before file access ('link following') in Windows iSCSI allows an unauthorized attacker to elevate privileges over a network.
-
CVE-2026-113605
Stirling-PDF from 2.13.2 through 3.2.0 contains an incomplete denylist in DatabaseService.validateSqlContent() that allows administrators to bypass H2 import restrictions using CREATE OR REPLACE ALIAS. Authenticated administrators can upload a crafted SQL backup to the import-database endpoint, where RUNSCRIPT defines H2 Java function aliases and restores Java-method execution.
-
CVE-2026-115915
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, ImageMediaIO passes attacker-controlled image bytes to Pillow through Image.open(BytesIO(data)) without restricting decoder selection, and the MIME label does not constrain the selected format. Image-capable request paths through parse_image() and fetch_image() can therefore dispatch an EPS payload to host components. A host with Ghostscript affected by CVE-2024-29510 can execute commands with the vLLM process's privileges, while environments using Pillow 12.0.0 through 12.2.0 can trigger CVE-2026-59203 and exhaust media-loading workers. Deployments without the applicable vulnerable dependency are not affected by that corresponding impact. This issue is fixed in version 0.32.0.
-
CVE-2026-123710
A flaw was found in Red Hat Quay. An unauthenticated client can obtain an anonymous registry token with an audience it controls, present that token to the service key rotation endpoint, and install its own approved signing key with no expiry. Registry tokens signed with that key are accepted for any repository, allowing read and write access to private images and replacement of repository tags.
-
CVE-2026-118746
Not failing securely ('failing open') in .NET allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-104068
JetBackup plugin for WordPress before 3.1.24.10 contains a symlink-based arbitrary file write vulnerability that allows an authenticated WordPress administrator to write attacker-controlled file contents outside the intended restore destination during backup import and restore. Attackers can craft a backup archive containing a symbolic link that archive extraction follows without validating targets or enforcing destination containment, potentially leading to server-side code execution depending on the writable destination and server configuration.
-
CVE-2026-124018
Gutenverse Form plugin for WordPress before 3.2.2 contains a CSV formula injection vulnerability that allows unauthenticated attackers to inject spreadsheet formula-trigger characters by submitting free-text field values to the public form-submit REST route, which sanitizes input with sanitize_text_field() without stripping characters such as =, +, @, or a leading minus. Attackers can store malicious formulas in entry-data post meta that are later exported unneutralized via fputcsv() or esc_html() in the CSV export functions, causing the formula to execute when an administrator opens the exported file, enabling data exfiltration to an attacker-controlled URL or command execution on the administrator's workstation.
-
CVE-2026-109411
An access control bypass flaw was identified in Samba's vfs_ceph_rgw module in it's openat() function. An authenticated user can bypass access controls to read or overwrite restricted files on shares configured with the experimental vfs_ceph_rgw module. When a client requests an existing file using mismatched character casing, the module fails to perform a case-insensitive lookup correctly, causing the server to treat the operation as creating a new file. As a result, the server skips standard permission checks on the existing file, allowing the attacker to access or overwrite sensitive data owned by other users.
-
CVE-2026-119988
Numeric truncation error in Windows SMB allows an authorized attacker to execute code over a network.
-
CVE-2026-59978
AdminNeo is a full-featured database management tool written in PHP. From 4.6.2 until 5.5.0, the unauthenticated `?script=version` endpoint accepts POST data without origin, CSRF or parameter-type validation. A `version[]` parameter becomes a PHP array that `bootstrap.inc.php` serializes into the shared `adminneo.version` cache. On later page loads, `design.inc.php` passes that cached value to `cookie()` and `rawurlencode()`, causing a `TypeError` rather than encoding a version string. A remote attacker can make the instance return HTTP 500 to all users for the cache's twenty-four-hour lifetime or until the poisoned cache is removed. Exploitation requires the web server to be able to write that shared cache; making it unwritable is an interim workaround. The issue retains the underlying type-validation defect reported previously in Adminer despite changes to AdminNeo's version-checking flow. This issue is fixed in version 5.5.0.
-
CVE-2026-116467
A security flaw has been discovered in ggml-org llama.cpp up to 0.7.0. Affected by this vulnerability is an unknown functionality of the file src/llama-sampler.cpp of the component Mirostat Sampler. The manipulation results in reachable assertion. It is possible to launch the attack remotely. The project was informed of the problem early through a pull request but has not reacted yet.
-
CVE-2026-122164
Weak authentication in Windows Kerberos allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-109476
Improper encoding or escaping of output in ASP.NET Core allows an unauthorized attacker to disclose information over a network.
-
CVE-2026-123776
RRD::CGI::Image versions through 0.01 for Perl pass CGI query parameters containting "tz" to tzset(), , so a FIFO or device path blocks the CGI process. The normalize_params method sends any query key matching /tz/i to tz(), which sets $ENV{TZ} to that value, with no validation. It then calls tzset(). glibc treats a TZ value starting with / or : as a file path and opens it, so a value naming a FIFO or a blocking device stalls the process inside tzset().
-
CVE-2026-113678
Froxlor before 2.3.15 contains a symlink following vulnerability in Statistics::makeChownWithNewStats that allows authenticated customers to take ownership of arbitrary host directories by abusing the root-run traffic cron. Attackers can replace the goaccess statistics directory in their document root with a symlink, causing chown -R to hand root-owned trees to them and enabling root escalation.
-
CVE-2026-97040
Building a package containing a crafted ".syso" file could bypass cgo flag validation and inject arbitrary linker flags when linking. When loading ".syso" entries from a package archive, the linker only checked for recognized host object headers (ELF, Mach-O, PE, and XCOFF) before falling through to Go object parsing. A ".syso" file beginning with a "go object" header was parsed as a Go compiler object, allowing untrusted cgo metadata directives (such as "cgo_ldflag" or "cgo_dynamic_linker") embedded in the ".syso" file to be passed to the external linker without validation. Additionally, ".syso" filenames longer than 16 bytes lost their ".syso" extension when truncated in archive headers, and unrecognized ".syso" formats forced external linking in pure-Go builds. The linker now rejects ".syso" files that are not recognized host object formats, and "cmd/go" and "cmd/pack" preserve the ".syso" suffix when truncating long archive member names.
-
CVE-2026-110789
libp2p is a JavaScript implementation of the libp2p networking stack. Prior to 5.1.22, the @libp2p/crypto code reached by the default Identify service accepts a remote peer’s crafted publicKey before peer-id or signature verification and passes it through publicKeyFromProtobuf to the ASN.1 DER decoder. readLength in packages/crypto/src/keys/rsa/der.ts accepts long-form element lengths larger than the remaining buffer, causing readInteger and related readers to loop synchronously over an attacker-claimed length from a very small key blob. Because decoding occurs in one event-loop tick before the remotePeer equality check in packages/protocol-identify/src/identify.ts, the Identify timeout cannot interrupt it, allowing repeated unauthenticated connections to stall all timers and peers and transiently exhaust memory. This issue is fixed in version 5.1.22.
-
CVE-2026-113825
Concurrent execution using shared resource with improper synchronization ('race condition') in Windows SMB Server Network Transport Driver (srvnet.sys) allows an authorized attacker to execute code over a network.
-
CVE-2026-115779
Effect is a TypeScript framework that consists of several packages that work together to help build TypeScript applications. From @effect/platform 0.1.0 until 0.97.1 and effect 4.0.0-beta.0 until 4.0.0-beta.103, the name, domain, and path fields passed to makeCookie() allow semicolon delimiters that serializeCookie() writes verbatim into Set-Cookie headers. A caller that supplies attacker-controlled values can append cookie attributes, including a broader Domain or, when the caller has not set those options itself, SameSite=None or Max-Age. Encoded cookie values and CR or LF characters remain unaffected. This can widen a cookie's scope or add otherwise-unset cookie attributes. This issue is fixed in @effect/platform 0.97.1 and effect 4.0.0-beta.103.
-
CVE-2026-69188
WeGIA is a web manager for charitable institutions. Prior to 3.9.1, a user with backup-management permission can upload a crafted gzip-compressed tar archive through web/html/configuracao/importar_dump.php, after which web/controle/control.php accepts a user-controlled modulo value and combines it with nomeClasse for require_once(). By using a phar:// path to the uploaded archive and selecting the whitelisted InternoControle, whose empty resource list is treated as authorized, the user can cause an embedded controller file and its requested method to execute as the web-server process. The attack path depends on the configured BKP_DIR, and enabling phar.readonly or phar.require_hash does not prevent PHP from reading the gzip tar archive through phar://. Successful exploitation permits access to or modification of sensitive data, configuration, credentials, source files, host commands, and services reachable by the web-server account. This issue is fixed in version 3.9.1.
-
CVE-2026-126081
Action Pack is a framework for handling and responding to web requests. Prior to 8.1.4.1 and 8.0.5.2, ActionController::Parameters#each_key delegates to the internal ActiveSupport::HashWithIndifferentAccess that holds the request parameters, so when called with a block it returns that hash instead of the Parameters instance. The hash does not respond to permitted?, so Active Model mass assignment (new, assign_attributes, update) accepts it without raising ActiveModel::ForbiddenAttributesError. An application that passes the return value of each_key to mass assignment could allow a remote, unauthenticated attacker to set attributes that the application does not permit, such as an authorization flag. This is an incomplete fix of CVE-2020-8164, which made each, each_pair and each_value return the Parameters instance but left each_key delegated. An application is affected if it passes the return value of ActionController::Parameters#each_key, called with a block, to mass assignment. Rails itself makes no such call. This issue is fixed in versions 8.1.4.1 and 8.0.5.2.
-
CVE-2026-126304
NLTK from 3.10.0 before 3.10.4 contains a symlink escape vulnerability in YCOECorpusReader that allows attackers to read files outside the corpus root by planting psd or pos directory symlinks. Attackers controlling a YCOE corpus can make parsed_sents() and tagged_words() return file content from other corpora or temporary files within the NLTK data sandbox.
-
CVE-2026-91141
A vulnerability in the default configuration of vsomeip allows a remote, unauthenticated attacker to trigger massive memory allocations by sending a small TCP packet with a large, attacker-declared SOME/IP length field. This can lead to immediate service termination via OOM-kill on memory-constrained systems or total host memory exhaustion through repeated connections.
-
CVE-2026-114156
Improper input validation in ASP.NET Core allows an unauthorized attacker to bypass a security feature over a network.
-
CVE-2026-113573
CAI Content Credentials is affected by an Improper Input Validation vulnerability that could result in a Security feature bypass. An attacker could leverage this vulnerability to bypass security measures and gain unauthorized write access. Exploit depends on conditions beyond the attacker's control. Exploitation of this issue does not require user interaction.