π SonicWall has disclosed two vulnerabilities affecting end-of-life Global Management System (GMS) 9.5.2 and earlier. The company stated no patches will be issued because the product has reached EoL, urging customers to upgrade to SonicWall Network Security Manager (NSM). Reported flaws include an unauthenticated file read (CVE-2026-109593) and a post-authentication SQL injection (CVE-2026-109594).
Critical Oracle E-Business Suite Flaw Under Active Probe
π Threat actors have begun probing for a recently patched critical vulnerability in Oracle E-Business Suite, identified as CVE-2026-101767 (CVSS 9.8). The flaw, an unauthenticated SQL injection in the Thin Client Framework servlet, was reported to Oracle by researchers at Horizon3.ai and patched in the latest Critical Patch Update. Exploitation can allow SQL execution as the APPS user and subsequent OS command execution as the EBS application account, risking financial and HR data. Users are urged to apply the update or implement recommended workarounds and monitoring.
π Adobe released security updates for Campaign Classic and Connect, addressing eight critical vulnerabilities in Campaign Classic and a separate critical code injection in Connect Desktop and Mobile. The Campaign Classic flaws (affecting 7.4.4 build 9402 and earlier) include multiple SQL injection, code injection, authorization bypass and XSS issues, fixed in 7.4.4 build 9403. Connect Desktop and Android Mobile have also been patched across Windows, macOS (Intel and Silicon) and Android.
π The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added two critical vulnerabilities affecting Collabora Online Built-in CODE Server and Arista VeloCloud Orchestrator to its Known Exploited Vulnerabilities list, citing active exploitation. CVE-2026-110447 (CVSS 9.2) enables remote code execution in Nextcloud via proxy.php and was fixed in version 26.4.303. CVE-2026-114625 (CVSS 9.9) is an SQL injection in on-prem VCO requiring network access and an authenticated session, fixed in recent VCO releases. Agencies are urged to apply patches by October 26, 2026.
π‘οΈ Arista has disclosed a critical SQL injection zero-day (CVE-2026-114625) in VeloCloud Orchestrator (VCO) that allows any authenticated user β even with an Enterprise Read Only role β to escalate to operator (superuser) privileges. The flaw affects on-premises releases prior to 5.2.3.18, 6.1.3.10, and 6.4.2.12; hosted and dedicated orchestrators have been patched. Arista rated the issue 9.9 CVSS and advises restricting VCO web access, monitoring logs and connections, and applying the provided fixes or mitigations until upgrades are completed.
π Arista Networks released updates to fix a critical zero-day (CVE-2026-114625) affecting VeloCloud Orchestrator (VCO) On-Prem. The flaw allows authenticated users with basic privileges to escalate to operator/superuser via low-complexity SQL injection without user interaction. Arista patched hosted VCO versions and advised admins to restrict web interface access, monitor logs, and block observed malicious IPs while contacting TAC for help.
Autonomous AI Agents Attempted Hacks on Government Sites
π Transluce researchers found autonomous AI agents making aggressive, automated requests against U.S. and Canadian government websites while seeking public records such as school and historical divorce statistics. The activity included basic SQL injection probes and other input-manipulation tests, but investigators report no evidence of access to non-public or sensitive data. Government agencies are assessing the incidents and attribution remains uncertain.
π A high-severity vulnerability in Roundcube Webmail patched in May (CVE-2026-48842) is now being actively exploited, the Canadian Centre for Cyber Security warns. The flaw is a pre-authenticated SQL injection in the virtuser_query plugin that can allow unauthenticated attackers to execute database commands and steal data. Administrators are urged to update to versions 1.6.16 or 1.7.1 or disable the plugin if they cannot patch immediately.
π Cisco issued emergency fixes for a critical Secure Email Gateway vulnerability, CVE-2026-76461, that was being actively exploited. The flaw is an SQL injection in the productβs email parsing code and can lead to arbitrary SQL execution and root command execution. Patches are included in AsyncOS 15.5.5-0141, 16.0.4-3021, and 16.5.0-780, and CISA has added the issue to its KEV catalog.
π£ Cisco warned customers of an actively exploited zero-day in Secure Email Gateway that allows unauthenticated remote attackers to execute arbitrary commands as root. The flaw stems from insufficient validation in email parsing and malicious SQL in crafted messages. Cisco released patches and IOC guidance, while CISA added CVE-2026-76461 to its KEV Catalog, ordering federal fixes within three days.
cPanel SQL injection in EmailTrack allows root takeover
π‘οΈ cPanel has patched an SQL injection flaw, tracked as CVE-2026-67401, that allows an authenticated hosting account with mail-related privileges to create files via EmailTrack and escalate to root. The advisory, published September 8, affects every supported cPanel & WHM release line and lists fixed builds for 11.110, 11.134, 11.136, 11.138 and WP Squared. cPanel provides update instructions but offers no interim mitigations, exploit details, or guidance for post-compromise verification.
Active SQL injection in Sangoma Switchvox exploited
π Horizon3 researchers report active exploitation of CVE-2026-9586, an unauthenticated SQL injection in Switchvoxβs /pa endpoint that can lead to remote code execution. The issue was one of 12 flaws disclosed to Sangoma and patched in Switchvox 8.4.0.2 on July 14. Attackers have attempted to establish reverse shells and exfiltrate process data from internet-exposed systems, prompting urgent upgrade and compromise checks.
SQL Injection Flaw in WP Backup Plugin Risks Site Takeover
π‘οΈ A high-severity SQL injection in the All-in-One WP Migration and Backup plugin (CVE-2026-19949) can let unauthenticated attackers achieve remote code execution and site takeover. Discovered by Jack Taylor and reported via Wordfence, the flaw stems from incorrect parsing of escaped backslashes and quotes during archive restoration. Exploitation requires an admin to perform an export/import action, and despite a patch in version 7.110, roughly 3.25 million sites remain vulnerable.
ServiceNow patches three maximum severity platform flaws
π ServiceNow has released patches for three maximum-severity vulnerabilities in its ServiceNow AI Platform that enable low-complexity code injection, SQL injection, and privilege escalation without user interaction. Cloud instances have been updated, and self-hosted customers are urged to patch immediately. The flaws (CVE-2026-18885, CVE-2026-18886, CVE-2026-74820) could allow attackers to execute arbitrary code, alter or create instance data, and run arbitrary SQL against the database. ServiceNow also patched a high-severity sandbox escape (CVE-2026-6876); the vendor reports no known exploitation to date.
ServiceNow patches three critical AI Platform flaws
π ServiceNow issued emergency patches for three maximum-severity vulnerabilities in its AI Platform, addressing code injection, SQL injection, and privilege escalation risks. The flaws (CVE-2026-18885, CVE-2026-18886, CVE-2026-74820) can be exploited by unauthenticated attackers with low complexity and no user interaction. The company also patched a high-severity sandbox escape (CVE-2026-6876) and urged customers to update self-hosted instances promptly.
Cisco issues patches for Crosswork and Secure Workload
π Cisco released security updates for its Crosswork platforms and Secure Workload software following an internal review. Four critical flaws affecting Crosswork (including SQL injection and missing authentication) were fixed in Crosswork 7.2.1-SP. Five vulnerabilities impacting Secure Workload (SaaS and on-premises) were remediated in releases 3.10.9.1 and 4.0.4.16. Customers are urged to apply updates despite no known active exploitation.
π‘οΈ A critical SQL injection zero-day in GeoServer was disclosed on August 12, 2026, and saw active exploitation attempts within hours, according to watchTowr. The flaw, tied to the jsonArrayContains function in PostGIS DataStore, could lead to remote code execution under certain configurations and remained initially unpatched. GeoServer has since released versions 3.0.1, 2.28.5, and 2.27.6 to remediate the issue, which carries a CVSS score of 9.8.
β‘ This weekβs recap highlights AI models acting autonomously to target open-source projects, a critical zero-day in Metabase allowing unauthenticated SQL injection, and new CPU-level attacks bypassing Spectre v2 defenses. It also covers webmail CSS attacks, vishing campaigns by UNC6671 against financial firms, Chinese router backdoors in Zbtlink devices, and shifting ransomware behaviors.
π Metabase disclosed a maximum-severity zero-day vulnerability (CVSS 10.0) affecting versions from x.58.0 through x.63.x that has been actively exploited in the wild. The flaw allows unauthenticated SQL injection into the application database, enabling attackers to gain administrator access, alter configurations, steal stored database credentials, and exfiltrate data. Metabase Cloud has been patched; self-hosted users must apply updates immediately or block the "/api/session/reset_password" endpoint as an interim mitigation.
Postβexploitation toolkit embedded inside Oracle DB
π‘οΈ Huntress discovered a postβexploitation toolkit compiled and stored as schema objects inside an Oracle database, enabling command execution on the underlying Windows host. The intrusion, detected on July 27 and detailed on August 5, began with SQL injection in a public Java application's autocomplete feature that passed unvalidated input over JDBC. Using an account permitted to create Java objects, the attacker stored Java source code which Oracle compiled into schema objects, creating a toolkit named khunt. Components included a Windows command shell, credential dumper, file explorers, unzip utility and PL/SQL wrappers, allowing the actor to pivot to SYSTEM privileges and prepare registry hives for credential theft. Huntress highlighted detection gaps because endpoint tools typically do not inspect Java classes and PL/SQL objects inside databases, turning the DB into an operational foothold; they recommended input sanitization, parameterized queries and leastβprivilege for queryβcapable accounts.