All news in category “Security Advisory and Patch Watch”

⚠️ RondoDox has been observed exploiting unpatched XWiki instances to weaponize a critical eval injection, CVE-2025-24893, enabling arbitrary remote code execution via the /bin/get/Main/SolrSearch endpoint. The flaw was patched in XWiki 15.10.11, 16.4.1 and 16.5.0RC1 in late February 2025, but scanning and exploitation surged in November, including botnet-driven DDoS and cryptocurrency miner deployments. Security vendors noted spikes in activity on November 7 and November 11 and observed RondoDox adding this vector on November 3, 2025. Administrators should apply vendor patches immediately and review logs and network traffic for indicators of compromise.

🚨 Fortinet confirmed a silent patch for a critical FortiWeb GUI path confusion zero-day (tracked as CVE-2025-64446) that is being "massively exploited in the wild." The flaw allowed unauthenticated HTTP(S) requests to execute administrative commands and create local admin accounts on internet-exposed devices. Fortinet released fixes in FortiWeb 8.0.2 (Oct 28) and later; administrators should upgrade, disable internet-facing management interfaces if they cannot update immediately, and audit logs for unauthorized accounts.

⚠️ Oligo Security researcher Avi Lumelsky disclosed a widespread insecure-deserialization pattern dubbed ShadowMQ that affects major AI inference engines including vLLM, NVIDIA TensorRT-LLM, Microsoft Sarathi-Serve, Modular Max Server and SGLang. The root cause is using ZeroMQ's recv_pyobj() to deserialize network input with Python's pickle, permitting remote arbitrary code execution. Patches vary: some projects fixed the issue, others remain partially addressed or unpatched, and mitigations include applying updates, removing exposed ZMQ sockets, and auditing code for unsafe deserialization.

🔒 Cybersecurity researchers disclosed a chain of remote code execution (RCE) vulnerabilities affecting AI inference frameworks from Meta, NVIDIA, Microsoft and open-source projects such as vLLM and SGLang. The flaws stem from reused code that called ZeroMQ’s recv-pyobj() and passed data directly into Python’s pickle.loads(), enabling unauthenticated RCE over exposed sockets. Vendors have released patches replacing unsafe pickle usage with JSON-based serialization and adding authentication and transport protections. Operators are urged to upgrade to patched releases and harden ZMQ channels, restrict network exposure, and avoid deserializing untrusted data.

⚠️ Fortinet has released an advisory for FortiWeb addressing CVE-2025-64446, a CWE-23 relative path traversal that can allow unauthenticated actors to execute administrative commands via crafted HTTP/HTTPS requests. Affected releases include multiple 7.x and 8.x versions; Fortinet provides specific upgrade targets (8.0.2+, 7.6.5+, 7.4.10+, 7.2.12+, 7.0.12+). If immediate upgrades are not possible, disable HTTP/HTTPS on internet-facing interfaces and, after remediation, review configurations and logs for unexpected modifications or unauthorized administrator accounts.

🔒 CISA has added CVE-2025-64446 — a Fortinet FortiWeb path traversal vulnerability — to its Known Exploited Vulnerabilities (KEV) Catalog after evidence of active exploitation. Under BOD 22-01, Federal Civilian Executive Branch agencies must remediate cataloged CVEs by the required due date. CISA strongly urges all organizations to prioritize timely patching, apply available mitigations, and monitor for indicators of compromise. CISA will continue to add vulnerabilities that meet catalog criteria.

⚠️ ASUS has released new firmware to patch a critical authentication bypass vulnerability tracked as CVE-2025-59367 that enables remote, unauthenticated attackers to log into vulnerable DSL routers exposed online. The update — firmware 1.1.2.3_1010 — addresses the issue for DSL-AC51, DSL-N16, and DSL-AC750. ASUS urges users to install the update immediately and, if they cannot, to disable Internet-facing services (remote access, port forwarding, DDNS, VPN server, DMZ, FTP) and use strong, unique passwords as temporary mitigations.

🚨 Researchers report an authentication bypass in Fortinet FortiWeb that is being actively exploited in the wild, allowing attackers to create privileged administrator accounts and fully compromise devices. watchTowr reproduced the issue, released a proof-of-concept and an artifact generator to help identify vulnerable appliances. The flaw is patched in FortiWeb 8.0.2, but Fortinet has not published a PSIRT advisory or assigned a CVE, and Rapid7 urges emergency patching for older versions.

⚠️ A path traversal vulnerability in Fortinet FortiWeb appliances is being actively exploited to create local administrative users without authentication. Researchers from Defused and PwnDefend described requests targeting the /api/v2.0/cmdb/system/admin%3f/../../../../../cgi-bin/fwbcgi endpoint that inject admin accounts. Rapid7 and others confirm versions 8.0.1 and earlier are affected, while 8.0.2 is believed to contain the fix. Administrators are urged to update immediately, review logs for fwbcgi access, and search for unexpected admin accounts.

🛡️ U.S. cybersecurity agencies warn that the Akira ransomware operation has expanded to encrypt Nutanix AHV virtual machine disk files, with the first confirmed incident in June 2025. Akira Linux encryptors have been observed targeting .qcow2 virtual disk files directly rather than using AHV management commands. The advisory cites exploitation of SonicWall CVE-2024-40766 and includes new IOCs and mitigation recommendations.

⚠️ ImunifyAV, a widely used Linux malware scanner, contains a remote code execution flaw in its AI-bolit component affecting versions prior to 32.7.4.0. The vulnerability is rooted in unsafe use of call_user_func_array during deobfuscation, which can execute attacker-supplied PHP function names when the scanner performs active unpacking. CloudLinux released fixes in late October and backported them on November 10; administrators should update to 32.7.4.0 or newer immediately to mitigate risk.

⚠️ Security researchers demonstrated that a rogue Model Context Protocol (MCP) server can inject JavaScript into the built-in browser of Cursor, an AI-powered code editor, replacing pages with attacker-controlled content to harvest credentials. The injected code can run without URL changes and may access session cookies. Because Cursor is a Visual Studio Code fork without the same integrity checks, MCP servers inherit IDE privileges, enabling broader workstation compromise.

⚠️ Quokka's assessment of the Uhale Android platform used in many consumer digital picture frames found devices that download and execute malware on boot. The tested units update to Uhale app 4.2.0, install a JAR/DEX payload from China-based servers, and persistently load it at every reboot. Devices were rooted, shipped with SELinux disabled and signed with AOSP test-keys, increasing exposure. Quokka disclosed 17 vulnerabilities (11 with CVEs) including remote code execution, command injection, an unauthenticated file server and insecure WebViews; researchers linked artifacts to Vo1d and Mezmess while the vendor did not respond to notifications.

🔒 Amazon Threat Intelligence disclosed an advanced APT campaign that weaponized zero-day vulnerabilities in Citrix NetScaler (Citrix Bleed 2, CVE-2025-5777) and Cisco Identity Services Engine (CVE-2025-20337). Attackers achieved pre-auth remote code execution via input-validation and deserialization flaws and deployed an in-memory web shell masquerading as the ISE IdentityAuditAction component. The implant registered as a Tomcat HTTP listener, used DES with nonstandard Base-64 encoding, required specific HTTP headers, and relied on Java reflection and bespoke decoding routines to evade detection.

🔒 CISA has ordered U.S. federal agencies to fully patch two actively exploited vulnerabilities in Cisco firewall appliances within 24 hours. Tracked as CVE-2025-20362 and CVE-2025-20333, the flaws permit unauthenticated access to restricted URL endpoints and remote code execution; chained together they can yield full device takeover. The agency emphasized applying the latest updates to all ASA and Firepower devices immediately, not just Internet-facing units.

⚠️ CISA published an advisory on November 13, 2025, warning that Brightpick AI devices — Mission Control and Internal Logic Control — contain multiple high-severity weaknesses that are remotely exploitable. Tracked as CVE-2025-64307, CVE-2025-64308, and CVE-2025-64309, the issues include missing authentication, hardcoded credentials in client-side JavaScript, and an unauthenticated WebSocket endpoint. Calculated scores reach up to CVSS v4 8.7, and CISA advises isolating affected systems, minimizing network exposure, and using secure remote access while conducting impact assessments.

⚠ Rockwell Automation reported a remotely exploitable vulnerability (CVE-2024-22019) in FactoryTalk Policy Manager that can lead to resource exhaustion and denial of service. The issue stems from Node.js HTTP handling of chunked transfer encoding (CWE-404) that permits unbounded reads from a single connection. Affected releases include Version 6.51.00 and earlier; Rockwell corrected the issue in Version 6.60.00. CISA assigns a high severity rating (CVSS v4 8.7) and recommends upgrading, minimizing network exposure, and isolating control networks behind firewalls.

🔒 AVEVA has released advisory ICSA-25-317-03 addressing a cryptographic weakness in AVEVA Edge (formerly InduSoft Web Studio) that could allow a local actor with read access to project or offline cache files to brute-force user or Active Directory passwords. The issue is tracked as CVE-2025-9317 and carries a CVSS v4 base score of 8.3. AVEVA provides a 2023 R2 P01 Security Update and recommends project migration, password resets, and tightened file access controls. This vulnerability is not remotely exploitable according to CISA.

⚠️ Rockwell Automation disclosed two local vulnerabilities in Studio 5000 Simulation Interface (version 2.02 and earlier) that allow path traversal–based local code execution (CVE-2025-11696) and a local SSRF that can trigger outbound SMB requests for NTLM hash capture (CVE-2025-11697). Both issues carry high severity (CVSS v4: 9.3 and 8.8) and are exploitable by low-complexity local attackers. Rockwell recommends upgrading to version 3.0.0 or later; CISA advises isolating control system networks, minimizing exposure, and following secure remote-access practices.

🔔 CISA released 18 Industrial Control Systems (ICS) advisories addressing security flaws across a broad set of vendors and product families. The advisories cover firmware, application software, and cloud services used in operational technology and industrial environments, including products from Siemens, Rockwell Automation, AVEVA, and Mitsubishi Electric. Administrators should review the advisories for technical details and apply vendor mitigations, patches, and compensating controls promptly to reduce risk to availability and safety.