CVE-2026-76461
A critical SQL injection vulnerability in Cisco Secure Email Gateway allows unauthenticated remote attackers to execute arbitrary commands with root privileges via crafted email messages.
Critical vulnerabilities, curated daily for security professionals
Sandbox escape flaws in patriksimek vm2 and maximum-severity issues across Microsoft cloud services lead yesterday's disclosures, with four vm2 CVEs scored 9.9 or 10 alongside CVE-2026-69843 in Microsoft Fabric and CVE-2026-62874 in Azure Billing at CVSS 10. Twenty-eight critical CVEs were disclosed, up 47% from the prior day's 19, while high-priority disclosures fell 29% to 62 for a total of 90. Also notable are CVE-2026-54734 in Prebid prebid-server-java at CVSS 10, CVE-2026-85885 in Microsoft 365 Copilot at 9.9, and CVE-2026-67100 in HCL BigFix Service Management at 9.8. The pattern favors code execution through untrusted input handling: JavaScript sandbox escapes, server-side ad exchange processing, and AI assistant integrations, all of which sit on paths that accept attacker-supplied content by design. Six CVEs carry confirmed active exploitation, including CVE-2026-76460 in Cisco Identity Services Engine (CVSS 10) and two JFrog Artifactory flaws, so verify version status with each vendor and restrict management interface access on identity, email gateway, and artifact repository systems first.
Immediate action: Prioritize the actively exploited systems: Cisco Identity Services Engine and Secure Email Gateway, JFrog Artifactory, Acronis Backup, and Google Pixel, followed by internet-reachable Microsoft cloud tenants and any Node.js application embedding vm2 for untrusted code execution. Confirm the fixed version and current fix status in each vendor's advisory before scheduling maintenance, since the affected version ranges vary by deployment and release channel. Where a fix is not yet deployable, restrict network access to management interfaces and review logs on the exploited products for signs of prior access.
CVSS score (e.g. 9.1) — severity from 0–10. Red marks critical (9+), orange high (7–8.9).
Exploitability — how hard the flaw is to attack, read from the CVSS vector:
The lower the bar on all three, the easier to exploit at scale — “Network · No privileges · No interaction” is the worst case: hit from anywhere, no credentials, no victim action.
🔴 Actively exploited — confirmed under attack in the wild (CISA’s Known Exploited Vulnerabilities catalog). Prioritize these regardless of score.
EPSS · Nth percentile — FIRST.org’s estimated chance a flaw is exploited within 30 days. We flag it only in the top 10% — a statistical signal it’s unusually likely to be targeted, separate from whether attacks are confirmed.
A critical SQL injection vulnerability in Cisco Secure Email Gateway allows unauthenticated remote attackers to execute arbitrary commands with root privileges via crafted email messages.
A logic error in the Google Pixel cellular modem component allows for unauthenticated, adjacent privilege escalation without user interaction.
Acronis Backup plugins for cPanel and Plesk contain an insecure file permissions vulnerability that allows authenticated users to perform local privilege escalation.
An authentication bypass vulnerability in the Cisco Identity Services Engine API allows unauthenticated, remote attackers to gain unauthorized access to the management interface.
An incorrect authorization vulnerability in JFrog Artifactory allows authenticated attackers to perform unauthorized actions, potentially leading to data compromise.
An improper authentication vulnerability in JFrog Artifactory allows unauthenticated users to obtain an internal token, potentially exposing sensitive resources even when anonymous access is disabled.
A sandbox escape vulnerability in vm2 allows authenticated attackers to execute arbitrary native code on the host system via the Node.js crypto module.
A path traversal vulnerability in GravitLauncher allows unauthenticated remote attackers to read sensitive files, including credentials and cryptographic keys, via specially crafted HTTP requests.
A sandbox escape vulnerability in vm2 allows authenticated attackers to execute arbitrary code on the host system via the node:test builtin module on Node.js 24 and newer.
An authentication bypass vulnerability in Microsoft Fabric allows unauthorized attackers to spoof identities and elevate privileges over the network.
HCL BigFix Service Management contains SQL injection and cross-tenant data exposure flaws, allowing attackers to extract sensitive system information and access unauthorized PII across organizations.
A vulnerability in Azure Billing involving insufficient verification of data authenticity allows unauthorized attackers to elevate privileges over the network.
Prebid Server Java is vulnerable to Server-Side Request Forgery due to improper validation of user-supplied parameters in bidder adapters, allowing unauthenticated attackers to reach internal services.
The vm2 sandbox library fails to properly isolate the host tls module, allowing unauthenticated attackers to manipulate the process-wide TLS trust store and intercept secure communications.
A command injection vulnerability in Microsoft 365 Copilot allows an authenticated attacker to elevate privileges over a network.
A sandbox escape vulnerability in vm2 allows authenticated users to achieve arbitrary native code execution in the host process by exploiting the node:sqlite module.
A sandbox escape vulnerability in vm2 3.11.6 allows unauthenticated attackers to execute arbitrary code on the host Node.js process via improper Promise rejection handling.
An unauthenticated sandbox escape in vm2 allows sandboxed code to access the host process's https.globalAgent, leading to credential theft and unauthorized request execution.
The vm2 sandbox library is vulnerable to remote code execution due to improper control of dynamically managed code resources when require.external is enabled.
The vm2 sandbox library fails to restrict access to os and dns builtins, allowing unauthenticated attackers to read host information and hijack global DNS resolution.
Migratico Lite is vulnerable to an unauthenticated remote code execution flaw in versions 2.6.8 and earlier, allowing attackers to execute arbitrary code on the underlying server.
HGiga OAKlouds-custom_page is vulnerable to insecure deserialization, allowing unauthenticated remote attackers to execute arbitrary code via maliciously crafted serialized content.
An unauthenticated authentication bypass vulnerability in OpenReception appointment-booking-software allows attackers to hijack staff sessions via malicious passkey injection.
A sandbox bypass in vm2 allows untrusted code to escape isolation by bypassing NodeVM builtin deny-list policies, leading to arbitrary host command execution.
An insufficient entropy flaw in Synology DiskStation Manager login logic allows unauthenticated remote attackers to perform arbitrary file read/write operations and cause denial of service.
An improper output encoding vulnerability in the SCGI component of Synology DSM allows unauthenticated remote attackers to perform arbitrary file read, write, and denial of service operations.
Chamilo LMS versions prior to 2.0.1 are susceptible to unauthenticated remote code execution, potentially allowing a full system compromise.
The vm2 sandbox library before 3.11.7 allows sandboxed code to access the host Node.js Buffer pool, enabling unauthorized memory disclosure and potential host-level read or write operations.
The vm2 sandbox library fails to protect host TypedArray and ArrayBuffer prototypes, allowing unauthenticated attackers to perform prototype pollution and achieve sandbox escape.
The vm2 sandbox library is vulnerable to a sandbox escape via the NodeVM component, allowing unauthenticated attackers to achieve arbitrary remote code execution on the host system.
The vm2 sandbox library contains a critical escape vulnerability allowing unauthenticated attackers to gain host Node.js capabilities through WebAssembly compilation streams.
The vm2 sandbox package fails to correctly validate package names during allowlist checks, allowing attackers to bypass security restrictions and execute unauthorized host packages.
A heap-based buffer overflow exists in the HappySeaFox sail library due to improper handling of TGA image headers, allowing unauthenticated attackers to achieve remote code execution via crafted files.
A heap-based buffer overflow in the HappySeaFox sail library allows unauthenticated attackers to cause memory corruption or remote code execution via a specially crafted PSD image file.
A vulnerability in the Apache Airflow Kafka provider allows authenticated users with connection-editing privileges to achieve arbitrary code execution in the Airflow scheduler process.
The FaustWP WordPress plugin fails to include the initialization vector in its HMAC authentication, allowing authenticated users to forge administrator access tokens and achieve remote code execution.
The ShortPixel Image Optimizer plugin for WordPress is vulnerable to PHP Object Injection via deserialization of untrusted input, potentially allowing code execution if a POP chain is present.
Disclosed Sep 13; published with a limited analysis after repeated re-checks found no further public detail.
The wpstorecart WordPress plugin through 5.0.7 is vulnerable to PHP object injection via an insecurely implemented deserialization process in a bundled add-on.
WeGIA versions prior to 3.8.5 contain an authorization bypass vulnerability allowing authenticated users to access, modify, or delete sensitive records belonging to other users.
Appwrite before 2.0.0 is vulnerable to argument injection, allowing authenticated users to execute arbitrary commands via the providerRootDirectory parameter.
The pgAdmin 4 Backup tool is vulnerable to argument injection and path traversal via the database field, allowing authenticated users to perform arbitrary file writes and credential theft.
The RabbitMQ Java client library is vulnerable to a denial of service attack via oversized frames, which can trigger memory exhaustion by bypassing configured message size limits.
An authenticated user in Curiosity Workspace can perform an unauthorized privilege escalation to become an administrator of an existing access group.
An authenticated path traversal vulnerability in Vvveb allows attackers to inject and execute arbitrary PHP code, potentially leading to persistent webshell placement and full system compromise.
The Mapster WP Maps plugin for WordPress is vulnerable to arbitrary user meta updates via the my_profile_update function due to a lack of proper input validation and capability checks.
The MCP Documentation Server incorrectly binds the unauthenticated document management API to all network interfaces, allowing remote attackers to access, modify, or delete sensitive documentation.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A buffer overflow vulnerability exists in the Linux kernel panasonic-laptop driver due to an off-by-one error when writing a trailing sentinel value to the sinf array.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A race condition in the Linux kernel core scheduling mechanism can lead to improper lock handling during task selection, potentially resulting in memory corruption or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A missing DMA mask initialization in the Linux kernel OMAP SSI driver can lead to system crashes or undefined behavior when performing DMA mapping operations.
The WPLP Cookie Consent WordPress plugin fails to sanitize user input in a public endpoint, leading to Stored Cross-Site Scripting (XSS) within administrative screens on multisite installations.
The Dewa Kirim WordPress plugin is vulnerable to stored cross-site scripting via unescaped delivery coordinates, allowing unauthenticated attackers to execute arbitrary JavaScript in administrator sessions.
The PuppyFW WordPress plugin lacks proper authorization on a REST route, allowing authenticated users to modify arbitrary blog options and escalate privileges.
An uncontrolled search path vulnerability in the NetBox Device Type Library allows unauthenticated contributors to achieve remote code execution on GitHub Actions runners via malicious module shadowing.
Disclosed Sep 12 without a CVSS score; scored Sep 13, analysis completed Sep 13.
The Zonify WordPress plugin fails to perform authentication checks, allowing unauthenticated attackers to retrieve stored account login tokens and hijack linked service accounts.
A symbolic link vulnerability in Grafana plugin extraction allows attackers to write arbitrary files and execute code on the host system.
SolarWinds Access Rights Manager contains a vulnerability involving a hardcoded static cryptographic key, which allows an unauthenticated attacker to achieve remote code execution.
A missing authorization flaw in AI Agent Automation allows authenticated attackers to access, delete, or clear memory belonging to other users, breaking tenant isolation.
A vulnerability in the LDAP API of Synology DiskStation Manager allows remote authenticated users to perform unauthorized file operations and denial of service attacks.
SiYuan versions before 3.8.4 contain a cross-site scripting vulnerability in the backlink dock tree due to improper escaping of document titles, allowing for potential remote code execution.
HUBzero CMS contains a path traversal vulnerability in project file upload handlers, allowing authenticated users to write arbitrary files to the server and potentially achieve code execution.
An unauthenticated path traversal vulnerability in Frappe LMS allows remote attackers to read arbitrary files from the server via the SCORMRenderer component.
An integer wraparound vulnerability in the ESPAsyncWebServer multipart/form-data parser allows unauthenticated remote attackers to trigger a denial of service via a crafted 256-byte boundary.
Umbraco CMS contains a security flaw in the Content Delivery API where authorization checks are bypassed for referenced nodes, allowing unauthorized access to sensitive content properties.
Improper validation of pod identifier uniqueness in the AWS Network Policy Agent can allow an authenticated user to bypass NetworkPolicy enforcement on co-located pods.
RabbitMQ amqp091-go fails to properly encode TLS asset paths in query strings, allowing for connection option injection when reparsing serialized URIs.
Opencast is vulnerable to session fixation via the JSESSIONID parameter, allowing an unauthenticated attacker to hijack authenticated user sessions, including those of administrators.
SiYuan fails to escape bookmark labels in notebook files, leading to Cross-site Scripting (XSS) that can execute arbitrary system commands via the Electron renderer.
The MobiAPParc password reset feature is vulnerable to unauthorized account modification due to improper validation of the user_id parameter, allowing unauthenticated attackers to reset user passwords.
A cross-site request forgery vulnerability in the Xagio SEO plugin allows unauthenticated attackers to perform unauthorized actions on behalf of users.
Admin3 versions 3.0.0 and earlier inadvertently include live user session tokens in audit log entries, allowing unauthorized privilege escalation by authenticated users with log viewing permissions.
A cross-site scripting (XSS) vulnerability exists in Paella Player, as used in Opencast, due to improper sanitization of closed captions cue text.
A memory safety vulnerability in OpenDDS allows an unauthenticated network attacker to cause a denial of service via a malformed RTPS UDP submessage.
A stored cross-site scripting (XSS) vulnerability in Vendure allows lower-privileged administrators to execute malicious scripts in the sessions of other administrators.
Pocketbase is vulnerable to a denial of service attack because its panic recovery middleware fails to handle internal worker goroutines, allowing an unauthenticated attacker to crash the server.
Advantech EKI-1242 series devices transmit sensitive management traffic in cleartext, allowing network-adjacent attackers to intercept sensitive device identity and network metadata.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A stack-based buffer overflow exists in the Linux kernel ACPI pfr_update driver, allowing local attackers to overwrite stack memory via malicious firmware-supplied buffer lengths.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A task_work cancellation vulnerability in the Linux kernel io_uring subsystem allows local attackers to potentially trigger improper waitid behavior due to incorrect context handling.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
The Linux kernel amd_pmc driver fails to clean up LPS0 handlers and debugfs entries during initialization errors, leading to kernel list corruption and potential system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A use-after-free vulnerability exists in the Linux kernel via-sdmmc driver due to improper interrupt handling during probe failure, which may lead to privilege escalation or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A use-after-free vulnerability in the Linux kernel via-sdmmc driver allows local attackers to trigger memory corruption and potential system compromise via a race condition during device removal.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A boundary error in the Linux kernel ISST driver allows local attackers to perform out of bounds memory access via the clos_assoc ioctl, potentially leading to privilege escalation or system crashes.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A race condition in the Linux kernel RDMA subsystem allows local users to trigger a use-after-free vulnerability, potentially leading to privilege escalation or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A use-after-free vulnerability exists in the Linux kernel RDMA/cxgb4 driver where improper cancellation of pending registration work leads to memory corruption during device removal.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A memory alignment validation flaw in the Linux kernel block layer allows local authenticated users to trigger potential data corruption or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A resource management flaw in the Linux kernel uvesafb driver fails to unregister a callback on initialization failure, potentially leading to local privilege escalation or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A use-after-free vulnerability in the Linux kernel fanotify subsystem allows local attackers with low privileges to trigger memory corruption or information disclosure.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A memory management flaw in the Linux kernel erofs filesystem allows for potential heap corruption or system instability when resizing global buffers during failed allocation attempts.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A memory management flaw in the Linux kernel eCryptfs subsystem allows a local user to cause a resource leak by failing to release a message context during send failures.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A buffer underflow vulnerability exists in the Linux kernel eCryptfs subsystem, where malformed tag 70 packets can trigger memory corruption during filename size calculation.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
The Linux kernel ecryptfs component fails to validate key sizes in Tag 3 packets, leading to heap-based buffer overflows during session key decryption.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A race condition in the Linux kernel eCryptfs subsystem allows for memory corruption due to improper locking during daemon queue management.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A memory safety vulnerability in the Linux kernel NTFS3 filesystem driver allows out-of-bounds memory access during log replay, potentially leading to system instability or code execution.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A memory leak in the Linux kernel Ceph filesystem driver during writeback aborts at unmount can lead to inode reference leaks, causing kernel panics and system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A flaw in the Linux kernel AMD PMC driver allows for improper handling of System Management Unit (SMU) command failures, potentially resulting in unauthorized memory access or system instability.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
An array index out of bounds vulnerability exists in the Linux kernel rtlwifi driver, specifically within the rtl8192du component, allowing for potential memory corruption.
Disclosed Sep 11 without a CVSS score; tracked by CVE Brief from Sep 12; scored Sep 13, analysis completed Sep 13.
A race condition in the Linux kernel NFS server (nfsd) allows for silent data loss during asynchronous copy operations by incorrectly sampling writeback error cursors.