Une faille Cisco ISE exploitée déclenche une vague d’attaques contre des canaux logiciels de confiance
La faille critique Cisco ISE CVE-2026-76460 est exploitée, avec d’autres risques touchant Orkes Conductor, libheif et les plug-ins d’IA.
Image d’illustration générée par IA
Les activités malveillantes observées le 21 septembre 2026 montrent que des attaquants transforment à répétition des infrastructures de confiance en vecteurs d’accès initial. Les API de gestion, les plug-ins d’agents d’IA, les extensions de navigateur, les scripts intégrés aux sites web et les services cloud légitimes sont tous impliqués dans les campagnes les plus récentes.
Le problème le plus urgent est CVE-2026-76460, une faille de contournement de l’authentification affectant Cisco Identity Services Engine, à laquelle un score CVSS de 10,0 a été attribué. Cisco a confirmé son exploitation active, tandis que la Cybersecurity and Infrastructure Security Agency a ajouté la vulnérabilité à son catalogue Known Exploited Vulnerabilities.
Parmi les autres risques immédiats figurent l’exécution de code sans authentification dans Orkes Conductor, une faille liée aux images malformées affectant libheif et Discourse, ainsi qu’une technique visant la chaîne d’approvisionnement des plug-ins d’IA et capable de remplacer silencieusement du code dont la version est verrouillée.
La faille de l’API Cisco ISE est déjà exploitée
CVE-2026-76460 affecte deux produits en version 3.1.0 :
- Cisco Identity Services Engine 3.1.0
- Cisco Identity Services Engine Passive Identity Connector 3.1.0
La vulnérabilité découle de contrôles d’authentification insuffisants sur un point de terminaison d’API. Un attaquant distant peut envoyer une requête spécialement conçue sans fournir d’identifiants, contourner l’authentification et obtenir un accès non autorisé via l’interface de gestion web.
Son vecteur CVSS 3.1 est CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H. Cette évaluation correspond à une attaque accessible par le réseau, de faible complexité, ne nécessitant ni privilèges ni interaction de l’utilisateur, et susceptible d’avoir de graves conséquences sur la confidentialité, l’intégrité et la disponibilité. La faiblesse est classée CWE-648.
La CISA a ajouté la faille au catalogue KEV le 16 septembre 2026 et fixé au 19 septembre 2026 la date limite de remédiation pour les agences fédérales américaines. Les mesures requises comprennent l’application des instructions d’atténuation de Cisco, des recommandations de gestion des correctifs fondées sur les risques de la directive BOD 26-04 et des exigences de triage forensique de la CISA. Les organisations doivent cesser d’utiliser le produit si aucune mesure d’atténuation efficace n’est disponible.
Les versions affectées sont connues, mais les informations disponibles n’identifient pas de version précise de Cisco dans laquelle la faille serait corrigée. Les administrateurs doivent donc suivre les instructions actuelles de l’éditeur plutôt que déduire une limite de version sûre.
L’exploitation ayant été confirmée, la réponse doit aller au-delà de l’application des correctifs. Les équipes de sécurité doivent examiner l’activité des API ISE, l’exposition de l’interface de gestion et les journaux d’authentification afin de repérer tout accès inhabituel. Les services de gestion exposés sur Internet ou largement accessibles doivent être traités en priorité.
Cette alerte KEV ne constitue pas un cas isolé parmi les éditeurs concernés. Au cours des 90 jours précédents, CVE-2025-39682, CVE-2025-39964, CVE-2026-53266, CVE-2026-87886, CVE-2026-76461 et CVE-2026-87491 ont également été ajoutées au catalogue pour des produits Linux, Google et Cisco. CVE-2026-76461, un autre identifiant Cisco, a été ajoutée le 14 septembre 2026.
Orkes et Discourse exposés à des voies distinctes d’exécution de code à distance
La nouvelle CVE-2026-58138 affecte Orkes Conductor avant la version 3.30.2, notamment la version 3.21.21. Elle affiche un score CVSS de 9,8 et permet à des attaquants non authentifiés d’exécuter des commandes du système d’exploitation via l’API de workflow.
Le problème concerne des évaluateurs GraalVM non isolés configurés avec HostAccess.ALL ou allowAllAccess(true). Avant même l’authentification, un attaquant peut soumettre des définitions de workflow intégrées contenant des expressions JavaScript ou Python malveillantes.
Les chemins d’exécution concernés incluent les types de tâches INLINE, LAMBDA, DO_WHILE et SWITCH. Des expressions malveillantes peuvent accéder à la réflexion Java ou lancer directement des sous-processus, transformant l’évaluation du workflow en exécution de commandes arbitraires.
Les déploiements d’Orkes Conductor doivent être mis à niveau vers la version 3.30.2 ou une version ultérieure. Les opérateurs doivent également restreindre l’accès non authentifié aux API de workflow et examiner les définitions soumises afin de repérer toute activité inhabituelle liée aux scripts, à la réflexion, aux évaluateurs ou aux sous-processus.
Une autre surface d’attaque apparaît avec CVE-2026-32882, une lecture hors limites d’un tampon du tas dans libheif, déclenchée par des images HEIF malformées. Les versions amont de libheif jusqu’à la 1.21.2 sont affectées, tandis que la version 1.22.0 est indiquée comme version amont corrigée.
L’erreur se produit dans HeifPixelImage::overlay(). Lorsqu’une image enfant de superposition utilise une profondeur de bits différente pour ses canaux alpha et couleur, la fonction indexe le plan alpha en utilisant le pas des canaux de couleur au lieu de alpha_stride. Dans une configuration documentée avec une image de 100 × 50, la lecture hors limites qui en résulte peut dépasser de 3 123 octets la limite du tampon alpha.
L’évaluation amont lui attribue un score CVSS de 7,1, mais l’avis de sécurité de Discourse évalue l’impact au niveau du produit à 8,8 et décrit une exécution de code à distance via l’importation d’images. Ces scores diffèrent, car ils correspondent à des contextes d’exploitation différents.
Les versions corrigées de Discourse sont 2026.7.0, 2026.6.1, 2026.5.2 et 2026.1.6. Les installations basées sur Docker peuvent être reconstruites selon la procédure habituelle :
./launcher rebuild app
Les versions prises en charge du cœur de Discourse isolent également le traitement des images lorsque le noyau sous-jacent le permet.
Hacktron a par ailleurs signalé l’exploitation combinée de la faille libheif et d’une mauvaise configuration SSO pour accéder aux comptes ChatGPT d’employés d’OpenAI et à des dépôts internes. Les problèmes signalés ont été corrigés 14 heures après leur divulgation responsable.
Plugin4Shell remet en cause la confiance accordée aux extensions d’agents d’IA dont la version est verrouillée
Plugin4Shell est une technique visant la chaîne d’approvisionnement et fonctionnant sans clic, qui affecte quatre environnements de développement assisté par IA :
- Claude Code
- OpenAI Codex
- GitHub Copilot
- Google Gemini CLI
L’attaque contourne la protection censée être assurée par l’épinglage du hachage des plug-ins. Un agent vérifie la référence de commit fournie par une marketplace, mais ne vérifie pas lui-même que le commit attendu a bien été récupéré depuis le dépôt.
Un attaquant contrôlant ce dépôt peut donc faire pointer la récupération vers du contenu malveillant alors que le SHA enregistré continue de paraître légitime. Le plug-in de remplacement est alors installé et exécuté sans qu’un clic de l’utilisateur soit nécessaire.
Les versions affectées et les limites entre versions corrigées n’ont pas été communiquées. Les utilisateurs doivent mettre à jour le logiciel agent en suivant les instructions de chaque fournisseur ; la validation côté marketplace ne peut pas à elle seule corriger une défaillance de vérification au sein du client.
Les entreprises doivent considérer les plug-ins comme des dépendances logicielles exécutables. La propriété des dépôts, les privilèges d’écriture et la provenance des plug-ins nécessitent les mêmes contrôles que ceux appliqués aux packages internes. La surveillance doit également couvrir les processus et les connexions réseau sortantes créés par les agents de codage, en particulier après l’installation ou la mise à jour d’un plug-in.
Les campagnes ClickFix passent par des services cloud et des scripts intégrés
Une compromission de la chaîne d’approvisionnement de Brevo montre comment un seul identifiant cloud peut exposer à la fois un fournisseur et ses clients. Le 14 septembre 2026, un attaquant a utilisé une clé d’API Brevo Cloudflare compromise pour déployer un Worker dans le compte de l’entreprise.
Pendant environ cinq heures et demie, le Worker a modifié des pages de brevo.com, des pages de sibforms.com et trois fichiers JavaScript intégrés par des clients. Plus de 100 000 sites web ont été exposés au code injecté.
Certains visiteurs ont vu apparaître un faux CAPTCHA Cloudflare leur demandant de coller puis d’exécuter une commande malveillante. Les administrateurs de sites pouvaient également recevoir un plug-in WordPress malveillant en consultant leurs propres sites.
Cet incident est distinct de la compromission SAML SSO antérieure de Brevo. Cette opération avait affecté 138 comptes, entraîné l’envoi de messages d’hameçonnage depuis six comptes et permis l’exportation de contacts depuis 43 comptes.
Une autre activité ClickFix s’est appuyée sur un Google Doc légitime pour cibler un chercheur en sécurité. Un Google Apps Script associé affichait une barre latérale personnalisée contenant une fausse erreur de déchiffrement et demandant de coller des commandes dans Terminal. La charge utile variait selon la plate-forme : voleur d’informations AMOS pour macOS et chaîne de chargement PowerShell pour Windows.
Une campagne de vol de cryptomonnaies a utilisé une autre variante. Elle récupérait du JavaScript obfusqué via l’API Google Visualization et tentait de convaincre les victimes de coller du code dans Chrome ou de l’installer avec Tampermonkey, qui assurait également la persistance.
Ces attaques ne reposent pas sur une vulnérabilité classique du navigateur. Leur point de contrôle commun est l’instruction donnée à l’utilisateur. Les processus légitimes de vérification CAPTCHA, de réparation de documents et de mise à jour logicielle ne demandent pas aux utilisateurs de coller du code arbitraire dans PowerShell, Terminal, une barre d’adresse ou la console d’une extension.
L’accès aux navigateurs et les attaques périphériques élargissent l’exposition
L’outil de malware bancaire KREMLIN installe des extensions malveillantes dans Google Chrome et Microsoft Edge. Active depuis au moins mai 2025 et suivie sous le nom REF9334, l’opération usurpe l’identité d’une douzaine de banques brésiliennes environ.
Ses chargeurs JavaScript en plusieurs étapes et ses installateurs C++ personnalisés déploient des extensions capables de dérober des identifiants, des jetons de session et d’autres données stockées dans le navigateur. Les victimes présumées doivent supprimer les extensions non autorisées, invalider les sessions de navigateur et renouveler les jetons.
Deux autres problèmes liés à KEV et aux plates-formes nécessitent une attention particulière. CVE-2026-58704, à laquelle un score CVSS de 8,8 a été attribué, est une faille de contournement des autorisations dans le modem cellulaire de Google Android. Elle nécessite un accès adjacent, mais ni privilèges ni interaction de l’utilisateur. La CISA l’a ajoutée au catalogue KEV le 16 septembre 2026 et fixé au 19 septembre la date limite de remédiation.
CVE-2026-82079 affecte les versions de Nintendo Switch antérieures à 23.0.0. Un attaquant à portée du réseau sans fil peut envoyer un trafic spécialement conçu au sous-système sans fil local et exploiter un dépassement de tampon sur la pile pour exécuter du code à l’aide d’une programmation orientée retour. La mise à jour vers la version 23.0.0 ou une version ultérieure corrige la plage de versions affectées documentée.
Enfin, CVE-2026-90894, connue sous le nom de ParaShells, permet une élévation locale de privilèges par l’intermédiaire du service prl_disp_service de Parallels Desktop, exécuté avec les privilèges root. En attendant la mise à disposition d’un correctif par l’éditeur, les organisations doivent limiter l’accès des utilisateurs locaux non fiables et surveiller toute installation inhabituelle d’appliances, toute exécution de tar ou tout processus enfant lancé avec les privilèges root.
La priorité opérationnelle est claire : corriger en premier les failles Cisco et Android répertoriées dans KEV, puis traiter l’exécution de workflows sans authentification, l’exposition liée au traitement des images et les canaux compromis de distribution de logiciels. Les interfaces de confiance accomplissent une grande partie du travail des attaquants.
Sources
Cet article est une réécriture originale fondée sur les sources ci-dessous.
- source primaireGitHub Security Advisory
- The Hacker News
CVE traitées dans cet article
- CVE-2026-85061Critique10.0MapLibre GL JS is an interactive vector tile map library for web browsers. Prior to 6.4.1, DOM.sanitize() in src/util/dom.ts iterates elem.attributes as a live NamedNodeMap while removeAttributes() removes attributes from the same collection, shifting indexes and skipping an adjacent dangerous attri
- CVE-2026-76460Critique10.0A vulnerability in an API of Cisco Identity Services Engine (ISE) could allow an unauthenticated, remote attacker to bypass authentication. This vulnerability is due to insufficient authentication control on an API endpoint. An attacker could exploit this vulnerability by sending a crafted reques
- CVE-2026-70416Critique10.0Dell ObjectScale, versions prior to 4.4.0.0, contains a Deserialization of Untrusted Data vulnerability. An unauthenticated attacker with remote access could potentially exploit this vulnerability, leading to Remote execution.
- CVE-2026-68488Critique9.9A Time-of-check Time-of-use (TOCTOU) race condition leading to insecure symlink following in Plesk causes local privilege escalation to root via arbitrary file/directory ownership takeover.
- CVE-2026-89049Critique9.9A server-side request forgery issue due to improper validation of equivalent address representations in the port forwarding to remote hosts functionality in Amazon AWS Systems Manager Agent (SSM Agent) before 3.3.4851.0 on all platforms might allow an authenticated remote user to bypass the remote d
- CVE-2026-91998Critique9.9Casdoor through 4.4.0 contains an authorization bypass vulnerability in the /api/mcp endpoint that allows attackers with any application's clientId and clientSecret to gain unrestricted access to user administration across all organizations. Attackers can enumerate user records including password sa
- CVE-2026-76672Critique9.9A vulnerability exists in the SD-WAN Orchestrator that may lead to the exposure of sensitive configuration information. An authenticated remote attacker with read-only privileges could exploit this vulnerability by sending a specially crafted request to the cache synchronization endpoint. Successful
- CVE-2026-76669Critique9.9Privilege escalation vulnerabilities exist in the API of HPE Networking EdgeConnect SD-WAN Orchestrator. Successful exploitation could allow a remote low-privileged authenticated user to escalate their privileges to those of an administrative user, leading to complete system compromise.
- CVE-2026-76670Critique9.9Privilege escalation vulnerabilities exist in the API of HPE Networking EdgeConnect SD-WAN Orchestrator. Successful exploitation could allow a remote low-privileged authenticated user to escalate their privileges to those of an administrative user, leading to complete system compromise.
- CVE-2025-39682Critique9.8In the Linux kernel, the following vulnerability has been resolved: tls: fix handling of zero-length records on the rx_list Each recvmsg() call must process either - only contiguous DATA records (any number of them) - one non-DATA record If the next record has different type than what has alrea
- CVE-2026-28323Critique9.8SolarWinds Web Help Desk is found to be affected by a SAML authentication bypass vulnerability. This requires the SAML 2.0 authentication method to be enabled.
- CVE-2026-65400Critique9.8An authentication issue was addressed with improved state management. This issue is fixed in macOS Golden Gate 27, macOS Sequoia 15.7.9, macOS Sonoma 14.8.9, macOS Tahoe 26.6.1, macOS Tahoe 26.7. An attacker on the network may be able to authenticate to Screen Sharing without valid credentials.
- CVE-2026-58138Critique9.8Orkes Conductor 3.21.21 before 3.30.2 contains an unauthenticated remote code execution vulnerability that allows remote attackers to execute arbitrary OS commands by submitting inline workflow definitions containing malicious JavaScript or Python expressions to the workflow API endpoint prior to au
- CVE-2026-80172Critique9.8Dell SCG 5.0 Appliance versions prior to 5.36.00.16 and Dell SCG 5.0 Application versions prior to 5.36.00.00, contains an Insufficient Verification of Data Authenticity vulnerability. An unauthenticated attacker with remote access could exploit this, leading to unauthorized access. This vulnerabili
- CVE-2026-82232Critique9.8Improper neutralization of special elements used in an SQL command ('SQL injection') vulnerability in Apache Syncope. An administrator with adequate entitlements can achieve execution of arbitrary SQL via stacked queries, leveraging unsanitized sort clauses for Task search. This issue affects Ap
- CVE-2026-76461Critique9.8A vulnerability in the email parsing of Cisco AsyncOS Software for Cisco Secure Email Gateway could allow an unauthenticated, remote attacker to execute arbitrary commands with root privileges on the underlying operating system. This vulnerability is due to insufficient validation in the email pa
- CVE-2026-65414Critique9.8An out-of-bounds write issue was addressed with improved bounds checking. This issue is fixed in iOS 26.7 and iPadOS 26.7, iOS 27 and iPadOS 27, macOS Golden Gate 27, macOS Sequoia 15.8, macOS Tahoe 26.7, tvOS 27, visionOS 27, watchOS 27. A remote attacker may be able to cause unexpected app termina
- CVE-2026-89026Critique9.8The Issabel Framework, the web framework supporting Issabel PBX software, before commit b97dbaf contains a hard-coded HS256 JWT signing key in the pbxapi index.php file that is identical across every installation, allowing unauthenticated remote attackers to forge valid bearer tokens. Attackers can
- CVE-2026-81642Critique9.8In NLnet Labs Unbound up to and including 1.26.0, a vulnerability was found in the DNSSEC validator that enables denial of service and possible remote code execution as a result of digesting DNSKEYs. A DNSKEY with an owner compression pointer to its own RDATA can overflow the digest buffer. Remote c
- CVE-2026-39919Critique9.8Ghostscript before 10.08.0 contains a heap-based buffer overflow vulnerability in the JPEG 2000 output adapter (base/sjpx_openjpeg.c) that allows attackers to cause memory corruption by supplying a crafted PDF containing a JPEG 2000 image with mismatched component subsampling factors. When image com
- CVE-2026-91843Critique9.8A stack overflow during the unauthenticated login process may allow an attacker to run arbitrary code remotely with root privileges.
- CVE-2026-90999Critique9.8Sentry Seer is vulnerable to a multi-stage trust-boundary violation that allows unauthenticated attacker-controlled telemetry to become code that is executed by an agent in a privileged automation environment. An external attacker can submit fabricated Sentry events without having access to the vict
- CVE-2026-76673Critique9.8Vulnerabilities have been identified in the API of EdgeConnect SD-WAN Orchestrator that could potentially allow an unauthenticated remote actor to circumvent existing authentication controls. Successful exploitation could allow an attacker to gain administrative privileges leading to complete compro
- CVE-2026-76674Critique9.8Buffer overflow vulnerabilities exist in the underlying operating system of HPE Networking EdgeConnect SD-WAN Gateways that could allow an unauthenticated remote attacker to execute arbitrary code. Successful exploitation could allow an attacker to execute arbitrary commands on the underlying operat
- CVE-2026-91749Critique9.6Use after free in Workers in Google Chrome prior to 153.0.8010.47 allowed a remote attacker to potentially execute arbitrary code outside the sandbox via a crafted HTML page. (Chromium security severity: Critical)
- CVE-2026-93372Critique9.6Buffer overflow in WebGL in Google Chrome on on Android prior to 153.0.8010.52 allowed a remote attacker to execute arbitrary code outside the sandbox via a crafted HTML page. (Chromium security severity: Critical)
- CVE-2026-93374Critique9.6Use after free in Dawn in Google Chrome on on Android prior to 153.0.8010.52 allowed a remote attacker to potentially execute arbitrary code outside the sandbox via a crafted HTML page. (Chromium security severity: Critical)
- CVE-2026-61410Critique9.4Dell SCG 5.0 Appliance versions prior to 5.36.00.16 and Dell SCG 5.0 Application versions prior to 5.36.00.00, contains a Missing Authorization vulnerability. An unauthenticated attacker with remote access could potentially exploit this vulnerability, leading to remote execution. This vulnerability
- CVE-2026-80238Critique9.3Dell SCG 5.0 Appliance versions prior to 5.36.00.16 and Dell SCG 5.0 Application versions prior to 5.36.00.00, contains an Execution with Unnecessary Privileges vulnerability. An unauthenticated attacker with local access could potentially exploit this vulnerability, leading to Protection mechanism
- CVE-2026-43790Critique9.1The issue was addressed with improved memory handling. This issue is fixed in macOS Golden Gate 27, macOS Sequoia 15.8, macOS Tahoe 26.7. A remote attacker may be able to cause unexpected system termination or corrupt kernel memory.
- CVE-2026-92034Critique9.1Site isolation issue in the Graphics component. This vulnerability was fixed in Firefox 156 and Thunderbird 156.
- CVE-2026-92038Critique9.1Mitigation bypass in the Remote Settings Client component. This vulnerability was fixed in Firefox 156, Firefox ESR 153.3, Thunderbird 156, and Thunderbird 153.3.
- CVE-2026-28309Critique9.1SolarWinds Serv-U is affected by a broken access control vulnerability that allows a domain administrator to create system administrator accounts. The impact is lower in Windows deployments.
- CVE-2026-28306Critique9.1SolarWinds Serv-U is affected by a privilege escalation vulnerability that allows a domain administrator to elevate their privileges to a system administrator. The impact is lower in Windows deployments.
- CVE-2026-28308Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to remote code execution. Domain administrator access is required. The impact is lower in Windows deployments.
- CVE-2026-28310Critique9.1SolarWinds Serv-U is affected by a privilege escalation vulnerability that allows a domain administrator to escalate their user type to that of a system administrator. The impact is lower in Windows deployments.
- CVE-2026-28314Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference vulnerability that leads to an account takeover. User authentication is required. The impact is lower in Windows deployments.
- CVE-2026-28313Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to SMTP hijacking leading to arbitrary account takeover. The impact is lower in Windows deployments.
- CVE-2026-28307Critique9.1SolarWinds Serv-U is affected by a privilege escalation vulnerability that allows a domain user group to be elevated into an administrator group. The impact is lower in Windows deployments.
- CVE-2026-28305Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to remote code execution as root. A domain account with admin privileges and read and write access to the home directory is required. The impact is lower in Windows deployments.
- CVE-2026-28317Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to privilege escalation. This issue requires domain administrator access. The impact is lower in Windows deployments.
- CVE-2026-28304Critique9.1SolarWinds Serv-U is affected by a remote code execution vulnerability that, when exploited, can allow the arbitrary execution of code remotely as root. The impact is lower in Windows deployments.
- CVE-2026-28312Critique9.1SolarWinds Serv-U is affected by a privilege escalation vulnerability. This would elevate a group’s access to system administrator and allow code execution as root. The impact is lower in Windows deployments.
- CVE-2026-28316Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to privilege escalation to a system administrator with the ability to execute commands as the root user. This issue requires a domain account with administrator access. The impact is lower in Wind
- CVE-2026-28302Critique9.1SolarWinds Serv-U is affected by an insecure direct object reference (IDOR) vulnerability that can lead to privilege escalation and remote code execution as root. This issue requires group administrator access. The impact is lower in Windows deployments.
- CVE-2026-28321Critique9.1SolarWinds Serv-U is affected by a broken access control vulnerability that could allow arbitrary file read and write, which can then be used to escalate privileges and execute code as root. A domain administrator access is required, and the impact is lower in Windows installations.
- CVE-2026-85982Critique9.0The Auth0 AD/LDAP Connector is vulnerable to stored Cross-Site Scripting (XSS) issues due to improper HTML encoding of data in search results and updater log content displayed in the admin panel. An authenticated user with privileges to modify directory attributes, or a low-privileged local user on
- CVE-2026-74469Élevée8.8In the Linux kernel, the following vulnerability has been resolved: sctp: prevent peer transport count overflow sctp_assoc_add_peer() increments the association's 16-bit transport_count for every new unique peer. Adding the 65,536th transport wraps the count to zero. SCTP sock_diag uses transport
- CVE-2026-65346Élevée8.8An integer overflow was addressed with improved input validation. This issue is fixed in iOS 26.6.1 and iPadOS 26.6.1, macOS Sequoia 15.8, macOS Tahoe 26.6.2, tvOS 27, visionOS 27, watchOS 27. Processing an image may lead to arbitrary code execution.
- CVE-2026-15315Élevée8.8Tapo C120 v1 and C200 v5 contain an improper authentication vulnerability within the login authentication verification module. An attacker on the local network can exploit weaknesses in challenge parameter validation to bypass normal authentication controls and obtain administrative session tokens.
- CVE-2026-87491Élevée8.8Out of bounds write in V8 in Google Chrome prior to 153.0.8010.36 allowed a remote attacker to execute arbitrary code inside the sandbox via a crafted HTML page. (Chromium security severity: Medium)
- CVE-2026-73693Élevée8.8FileRun before 2026.3.0 contains an OS command injection vulnerability in the PhotoProofSheet handler that allows authenticated users with upload permission to execute arbitrary commands by uploading files with shell metacharacters in their names. Attackers can upload a file containing command subst
- CVE-2026-78175Élevée8.8The Tutor LMS – eLearning and online course solution plugin for WordPress is vulnerable to PHP Object Injection in all versions up to, and including, 4.0.7 via the `withdraw_method_field` parameter of the `tutor_save_withdraw_account` AJAX handler. This is due to the handler lacking any capability o
- CVE-2026-58704Élevée8.8In Cellular Modem, there is a possible permission bypass due to a logic error in the code. This could lead to remote (proximal/adjacent) escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.
- CVE-2026-53266Élevée8.8In the Linux kernel, the following vulnerability has been resolved: netfilter: bridge: make ebt_snat ARP rewrite writable The ebtables SNAT target keeps the Ethernet source address rewrite behind skb_ensure_writable(skb, 0). This is intentional: at the bridge ebtables hooks the Ethernet header is
- CVE-2026-92013Élevée8.8Privilege escalation due to incorrect boundary conditions in the Graphics: CanvasWebGL component. This vulnerability was fixed in Firefox 156, Firefox ESR 115.41, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-92015Élevée8.8Privilege escalation in the WebExtensions component. This vulnerability was fixed in Firefox 156, Firefox ESR 115.41, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-92020Élevée8.8Privilege escalation due to incorrect boundary conditions in the Graphics: WebRender component. This vulnerability was fixed in Firefox 156, Firefox ESR 115.41, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-92022Élevée8.8Use-after-free in the DOM: HTML Parser component. This vulnerability was fixed in Firefox 156, Firefox ESR 115.41, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-92029Élevée8.8Use-after-free in the SVG component. This vulnerability was fixed in Firefox 156, Firefox ESR 115.41, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-92033Élevée8.8Privilege escalation in Firefox for Android. This vulnerability was fixed in Firefox 156.
- CVE-2026-91721Élevée8.8Use after free in Internals in Google Chrome prior to 153.0.8010.47 allowed a remote attacker to potentially execute arbitrary code outside the sandbox via a crafted HTML page. (Chromium security severity: Critical)
- CVE-2026-28326Élevée8.8SolarWinds Access Rights Manager was reported to be affected by an unauthenticated remote code execution vulnerability. The issue stems from a hardcoded static key.
- CVE-2026-18851Élevée8.8Missing authorization in Ivanti Endpoint Manager Mobile before version 12.10.0.0, 12.9.0.2, and 12.8.0.4 allows a remote authenticated attacker to escalate their privileges to admin.
- CVE-2026-91931Élevée8.5Flowise before 3.1.4 contains a remote code execution vulnerability in the Custom MCP node that allows authenticated attackers to execute arbitrary code by supplying npx package names in the mcpServerConfig parameter. Attackers can invoke npx with attacker-controlled npm packages to execute code on
- CVE-2026-91932Élevée8.5Flowise before 3.1.4 contains a validation bypass vulnerability in MCP server configuration allowing authenticated attackers remote code execution through an unvalidated cwd parameter. Attackers can bypass path validation using clean filenames in the args array while controlling the working director
- CVE-2026-82079Élevée8.4A stack-based buffer overflow vulnerability in the Nintendo Switch local wireless networking functionality may allow an attacker within wireless range to execute arbitrary code using return-oriented programming (ROP) through crafted network traffic. This issue affects Nintendo Switch: before 23.0.0.
- CVE-2026-78626Élevée8.1The Okta Access Gateway improperly handles input sanitization and regular expression evaluation within its Protected Rule authorization check, resulting in an authorization bypass when an administrator has explicitly configured a Protected Rule policy on one or more application resources.
- CVE-2025-43936Élevée8.1Dell ObjectScale, versions prior to ObjectScale 4.4.0.0, contains an Improper Authentication vulnerability. An unauthenticated attacker with remote access could potentially exploit this vulnerability, leading to Unauthorized access.
- CVE-2025-39964Élevée7.8In the Linux kernel, the following vulnerability has been resolved: crypto: af_alg - Disallow concurrent writes in af_alg_sendmsg Issuing two writes to the same af_alg socket is bogus as the data will be interleaved in an unpredictable fashion. Furthermore, concurrent writes may create inconsiste
- CVE-2026-68121Élevée7.8In the Linux kernel, the following vulnerability has been resolved: pppoe: reload header pointer after dev_hard_header() pppoe_sendmsg() saves a pointer to the PPPoE header before calling dev_hard_header(). Device header callbacks are allowed to reallocate the skb head, invalidating pointers into
- CVE-2026-81000Élevée7.8In the Linux kernel, the following vulnerability has been resolved: net: tun: bound receive headroom tun_get_user() uses tun->align both as skb headroom and when choosing how much packet data to keep linear. OVS can propagate an oversized headroom request from another port to TUN or TAP. When ali
- CVE-2026-90894Élevée7.8Parallels Desktop runs prl_disp_service as root. Local clients reach it on the world-writable socket /var/run/prl_disp_service.socket. PrlSrv_LoginLocal accepts peer credentials. No Parallels signature. No admin group. After login, PrlSrv_InstallAppliance lets you pick the appliance folder (sVmPa
- CVE-2026-43502Élevée7.8In the Linux kernel, the following vulnerability has been resolved: net/rds: handle zerocopy send cleanup before the message is queued A zerocopy send can fail after user pages have been pinned but before the message is attached to the sending socket. The purge path currently infers zerocopy stat
- CVE-2026-84607Élevée7.8A race condition was addressed with improved state management. This issue is fixed in iOS 26.7 and iPadOS 26.7, iOS 27 and iPadOS 27, macOS Golden Gate 27, macOS Sequoia 15.8, macOS Tahoe 26.7, tvOS 27, visionOS 27, watchOS 27. A sandboxed app may be able to execute arbitrary code with kernel privil
- CVE-2026-87886Élevée7.8Local privilege escalation due to insecure file permissions. The following products are affected: Acronis Backup plugin for cPanel & WHM (Linux) before build 1.9.3.1021, Acronis Backup extension for Plesk (Linux) before build 1.8.11.638, Acronis Backup plugin for DirectAdmin (Linux) before build 1.2
- CVE-2026-78623Élevée7.7The Okta Access Gateway does not sanitize SAML assertion values before interpolating them into database queries in the advanced mode datastore configuration. The unsanitized values are substituted directly into the query string prior to preparation, resulting in unintended SQL execution against the
- CVE-2026-73178Élevée7.5Exposure of Sensitive Information to an Unauthorized Actor vulnerability in Apache Syncope. An administrator with adequate entitlements can get access via REST to the list of existing Access Tokens, including their signed JWT body. These values can be then used to perform further REST requests, imp
- CVE-2026-19667Élevée7.5If an attacker-controlled authoritative server can produce a negative answer that is exactly 65536 bytes, then a flaw in `named` results in a negative cache entry of 0 bytes. When this entry is subsequently read, `named` aborts. This issue affects BIND 9 versions 9.11.0 through 9.18.50, 9.20.0 throu
- CVE-2026-77692Élevée7.5An attacker can cause `named` to abort by sending a crafted DNS-over-HTTPS request with a cryptographically invalid SIG(0) record, and then closing the transport connection prematurely. This issue affects BIND 9 versions 9.20.0 through 9.20.27, 9.21.0 through 9.21.25, and 9.20.9-S1 through 9.20.27-S
- CVE-2026-19666Élevée7.5On a resolver configured to use ``dns64``, if an applicable answer from the authoritative server is malformed in a specific way, the resolver `named` process will exit unexpectedly. This issue affects BIND 9 versions 9.11.0 through 9.18.50, 9.20.0 through 9.20.27, 9.21.0 through 9.21.25, 9.11.3-S1 t
- CVE-2026-76163Élevée7.5If BIND is loaded with a "`named.conf`" file that contains no global "`options`" block, an attacker can send a query of QTYPE TKEY which may cause an assertion failure and subsequent unexpected program exit. This issue affects BIND 9 versions 9.20.0 through 9.20.27, 9.21.0 through 9.21.25, and 9.20.
- CVE-2026-80274Élevée7.5If a BIND resolver sends a query for a DNSSEC-signed authoritative zone, and the authoritative server replies with a valid wildcard answer and signed NSEC3 proof, followed by an unsigned NSEC at the same owner name, it will trigger an unexpected program exit. This issue affects BIND 9 versions 9.11.
- CVE-2026-73694Élevée7.2FileRun before 2026.3.0 contains an OS command injection vulnerability caused by a no-op redefinition of escapeshellcmd() in CLI.php that strips shell-metacharacter escaping, allowing attacker-controlled input to reach an exec() sink unsanitized. Attackers can exploit this through an interactive pat
- CVE-2026-73698Élevée7.2FileRun before 2026.3.0 contains a SQL injection vulnerability that allows delegated or simple administrators to execute arbitrary SQL by submitting the description parameter as an array, causing the getValuesString() method in DB/DP.php to interpolate raw array values directly into an INSERT statem
- CVE-2026-73699Élevée7.2FileRun before 2026.3.0 contains a PHP object injection vulnerability that allows authenticated attackers to execute arbitrary code by exploiting incorrect options passed to unserialize() in the Perms::getPerms() method, where a positional array is used instead of the required named-key array to dis
- CVE-2026-32882Élevée7.1libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and prior contain a heap buffer over-read in HeifPixelImage::overlay() in libheif/pixelimage.cc. When compositing an overlay image (iovl) whose child image has a different bit depth for the alpha channel than for the color c
- CVE-2026-56711Élevée7.0VLC media player versions 3.0.0 through 3.0.23 contain a memory-safety vulnerability reachable when processing crafted media. Exploitation requires user interaction and may result in application termination or code execution with the privileges of the VLC process.
- CVE-2026-65812Moyenne6.8Insertion of sensitive information into sent data in Microsoft Teams for Android allows an authorized attacker to disclose information over a network.
- CVE-2026-15316Moyenne6.5An improper input validation vulnerability in the configuration service for processing encrypted credential data has been identified in Tapo C200 v5. An attacker can send oversized crypted ciphertext values that may trigger exception handling failures, due to insufficient validation, causing the af
- CVE-2026-77147Moyenne6.5Improper Control of Generation of Code ('Code Injection') vulnerability in Apache Syncope. An administrator with adequate entitlements for Implementations can create a malicious Groovy Command class containing untrusted code in their CommandArgs static implementation, bypassing the Groovy security
- CVE-2026-28315Moyenne6.2SolarWinds Serv-U was found to be affected by a stored cross-site scripting vulnerability that could lead to session hijacking or information disclosure from an administrator account.
- CVE-2026-92005Moyenne5.3Use-after-free in the Audio/Video: Web Codecs component. This vulnerability was fixed in Firefox 156, Firefox ESR 140.16, Firefox ESR 153.3, Thunderbird 156, Thunderbird 140.16, and Thunderbird 153.3.
- CVE-2026-91726Moyenne4.7Out of bounds read in WebGL in Google Chrome on on Android prior to 153.0.8010.47 allowed a remote attacker to read memory outside the sandbox via a crafted HTML page. (Chromium security severity: Critical)
- CVE-2026-73324Moyenne4.3Certain VLC media player builds in versions 3.0.0 through 3.0.23 contain a memory-safety vulnerability reachable when processing media from an attacker-controlled network source. Exploitation requires user interaction and may disclose a limited, layout-dependent amount of VLC process memory. Exposur
- CVE-2026-80844In the Linux kernel, the following vulnerability has been resolved: xfrm: ah6: validate routing header segments_left AH6 rearranges routing-header addresses before computing or verifying the ICV. ipv6_rearrange_rthdr() assumes that segments_left is not larger than the number of addresses described
- CVE-2026-0310A buffer overflow vulnerability in the XML processing functionality of Palo Alto Networks PAN-OS® software enables an unauthenticated attacker with network access to the management web or dataplane interface to cause a denial of service (DoS) condition on VM-Series firewalls or execute arbitrary cod
- CVE-2026-65638Improper escaping of a request URL in ConfigServer Security & Firewall allows an unauthenticated remote attacker to execute arbitrary commands as the CSF service account via shell command injection. The vulnerability affects versions of the software originally distributed by ConfigServer, as well
- CVE-2026-82717In NLnet Labs Unbound up to and including 1.26.0, a vulnerability was found in that can progressively corrupt heap memory and under certain systems and compilation options could lead to remote code execution. The vulnerability starts when CNAME synthesis during an upstream response needs to enforce(
- CVE-2026-77179On macOS, the virtio-fs host server used by Docker Sandboxes improperly follows symlinks when reopening an unlinked file from a stored path. A malicious guest can replace a parent directory with a symlink, escape the shared workspace, and read or modify arbitrary host files as the VMM user, potentia
