Generated · Archived snapshot 18 findings · 9 CVEs
On this page

Exploitation Report

Executive Summary

Active exploitation of critical zero-day and recently disclosed vulnerabilities has intensified across multiple vendor platforms in September 2026. Chinese threat actor UTA0565 chained three zero-days across Google Chrome and Microsoft Windows to deploy CLEANGULP malware, while separate campaigns actively exploit pre-authentication RCE flaws in Check Point Security Gateway VPN (CVE-2026-85102), F5 BIG-IP APM OAuth servers (CVE-2026-94127), and WordPress (CVE-2026-87902).

A MikroTik RouterOS SSH vulnerability chain (CVE-2026-67279 + CVE-2026-86060) enables passwordless administrative takeover of internet-exposed devices. Simultaneously, supply chain attacks have compromised the HashiCorp Registry with malicious Terraform providers, the npm/PyPI ecosystems with trojanized MemTensor packages, and GitLab's email-based issue submission workflow. An unpatched Ubuntu container escape (CVE-2026-80521) has a public exploit while vendor patches lag upstream fixes.

Active Exploitation Details

CriticalActive exploitationPatchCVE-2026-85102#

  • Description: Pre-authentication remote code execution vulnerability in the VPN certificate-handling functionality of Check Point Security Gateway products.
  • Impact: Unauthenticated attackers can achieve remote code execution on affected Security Gateway appliances.
  • Status: Actively exploited in the wild; Check Point has confirmed exploitation and released advisories.

CriticalActive exploitationPatchCVE-2026-87902#

  • Description: Critical vulnerability in WordPress allowing attackers to write executable files to disk that run shell commands when accessed.
  • Impact: Remote code execution on vulnerable WordPress sites; threat actors have moved from probing to active exploitation.
  • Status: Active exploitation confirmed; patches available.

CriticalActive exploitationPatchCVE-2026-67279CVE-2026-86060#

  • Description: Two chained vulnerabilities in MikroTik RouterOS SSH: an SSH state-machine flaw (CVE-2026-67279) combined with an argument-injection bug in the login process (CVE-2026-86060). The chain allows full administrative control without password, SSH key, or completed authentication.
  • Impact: Complete takeover of internet-exposed MikroTik routers; attack logs date back to earlier activity.
  • Status: Actively exploited in the wild; CERT Polska tracks the chain as "MikroTrick."

HighObservedMitigateCVE-2026-80521#

  • Description: Use-after-free in the Linux kernel's AF_UNIX socket subsystem enabling container escape to host root. Fixed upstream on August 6, 2026, but Ubuntu has not shipped patches for 26.04, 24.04, or 22.04 LTS releases.
  • Impact: Attackers with container access can escape to host and gain root privileges.
  • Status: Public exploit released; Ubuntu LTS releases remain unpatched despite upstream fix.

CriticalActive exploitationPatchCVE-2026-94127#

  • Description: Critical zero-day in F5 BIG-IP Access Policy Manager (APM) when configured as an OAuth authorization server. Allows unauthenticated remote code execution on the BIG-IP system.
  • Impact: Unauthenticated RCE on BIG-IP systems serving as OAuth servers; actively exploited before patch release.
  • Status: Actively exploited; F5 released engineering hotfixes on September 22, 2026.

CriticalActive exploitationPatchCVE-2026-85046CVE-2026-87491CVE-2026-85880#

  • Description: Chinese threat actor UTA0565 exploited a chain of three zero-days—two in Google Chrome (CVE-2026-85046, CVE-2026-87491) and one in Windows Advanced Local Procedure Call (CVE-2026-85880)—via fake websites to deploy CLEANGULP malware.
  • Impact: Full compromise via browser-to-kernel exploit chain; malware deployment on targeted systems.
  • Status: Exploited as zero-days on September 3–4, 2026; patches presumably underway or released by vendors.

CriticalActive exploitationPatch#

  • Description: Zero-day vulnerability in VeloCloud Orchestrator (VCO) On-Prem deployments actively exploited before patch availability.
  • Impact: Compromise of VCO management infrastructure; details limited in public reporting.
  • Status: Actively exploited; Arista released security patches on September 22, 2026.

CriticalPotentialPatch#

  • Description: Flaw in cPanel's CalDAV and CardDAV service allowing any hosting account holder to execute code as root and take full server control. A second bug in the WP Toolkit plugin permits cross-account database manipulation.
  • Impact: Complete server compromise from a standard hosting account; lateral movement across hosted accounts.
  • Status: cPanel released fixed versions for both issues on September 22, 2026; exploitation status not explicitly confirmed but high risk.

CriticalNot observedPatch#

  • Description: Vulnerability in Next.js ImageResponse feature allowing server code execution via crafted SVG input when attacker-controlled values (e.g., URL parameters) are rendered into images.
  • Impact: Remote code execution on Next.js applications using ImageResponse with user-supplied input.
  • Status: Fixed by Vercel in version released September 22, 2026; no confirmed exploitation reported.

HighPotentialMitigate#

  • Description: Automatically assigned incoming email addresses in GitLab contain highly privileged access tokens. Leaked addresses allow attackers to push code and run CI/CD jobs as the victim user, including to protected branches.
  • Impact: Supply chain compromise via credential theft; unauthorized code commits and CI/CD execution with victim's permissions.
  • Status: Design flaw with active risk; no patch indicated—mitigation requires treating the email address as a secret.

HighObservedInvestigate#

  • Description: First observed use of HashiCorp's centralized Terraform Registry to distribute Go-based malware via two malicious Terraform providers and two Go modules (e.g., gocommunity-io/dockerd, kreuzwenker/...).
  • Impact: Supply chain compromise of infrastructure-as-code pipelines; malware execution during Terraform runs.
  • Status: Disclosed by Aikido; packages identified with download counts (e.g., 222 downloads for dockerd); no CVE assigned.

HighObservedInvestigate#

  • Description: Legitimate MemTensor packages on npm and PyPI compromised to deliver sckit, a cross-platform Go-based credential stealer for Windows, Linux, and macOS.
  • Impact: Credential theft across developer environments; platform-specific implants deployed via trusted package managers.
  • Status: Reported by Aikido, SafeDep, Socket, and StepSecurity; affected versions identified.

HighPotentialMitigate#

  • Description: Technique injecting code into process initialization structures without using Windows APIs monitored by EDR tools, evading detection during process injection.
  • Impact: Bypass of endpoint detection and response controls; stealthy code execution.
  • Status: Technique disclosed; no CVE assigned; exploitation potential high against unpatched EDR configurations.

HighActive exploitationInvestigate#

  • Description: Financially motivated threat actor using open-source AI agent frameworks to attack hundreds of online retailers at scale, injecting skimmers and stealing over 600,000 credit card records.
  • Impact: Mass payment card theft; automated compromise of e-commerce platforms.
  • Status: Active campaign observed; infrastructure and tactics documented.

CriticalActive exploitationInvestigate#

  • Description: InfraTrust report indicates attackers increasingly targeting enterprise infrastructure management systems, with several critical vulnerabilities exploited before or shortly after vendor disclosure.
  • Impact: Potential full control of network infrastructure, lateral movement, persistence.
  • Status: Active exploitation of multiple undisclosed vulnerabilities; specific CVEs not enumerated in reporting.

HighPotentialMitigate#

  • Description: Confused deputy problem in Google Kubernetes Config Connector allows a limited-privilege Kubernetes user to escalate to organization-wide control via a single YAML file.
  • Impact: Full GCP organization takeover from compromised cluster workload identity.
  • Status: Research disclosure by Varonis; no active exploitation reported; mitigation requires IAM hardening.

MediumPotentialMonitor#

  • Description: Attack allowing privileged actors to register a rogue external MFA provider that intercepts user passwords during legitimate login flows.
  • Impact: Credential harvesting bypassing MFA protections; requires initial privileged access.
  • Status: Proof-of-concept demonstrated by researchers; no confirmed wild exploitation.

MediumNot observedMonitor#

  • Description: Windows malware (CLOSEDQUORUM) using up to four AI models in a voting mechanism to decide malicious actions (credential theft, browser password extraction, crypto wallet targeting) instead of traditional C2.
  • Impact: Autonomous, resilient malware decision-making; potential evasion of static C2 analysis.
  • Status: Public version non-functional; Talos has not observed full operational capability.

Affected Systems and Products

Attack Vectors and Techniques

Threat Actor Activities