Threat Intelligence Brief - Friday, October 23, 2026

Operational threat reporting for defenders who need signal, not noise.

By DevSecOpsDad
Posted on October 23, 2026

Threat Radar

  • A Chinese state-sponsored APT is actively exploiting ConnectWise using an AI-assisted penetration testing tool called Artex, achieving privileged access across hundreds of organizations — MSPs and their downstream customers are at immediate risk.

  • VMware VeloCloud Orchestrator has been hit by its third exploited zero-day in four months, a critical SQL injection flaw granting operator-level privileges — sustained attacker focus on this product signals it is a high-value target for network infrastructure compromise.

  • PHP disclosed a second unpatched RCE vulnerability (CVE-2026-115560) within 48 hours of emergency PHP2Shell patches — the rapid cadence of critical PHP disclosures is drawing attacker attention to internet-facing PHP applications.

  • A critical RCE in Python’s Pillow image library (CVE-2026-112900) allows code execution via crafted EPS files processed through Ghostscript — any Python application accepting untrusted image uploads is potentially exposed.

  • Sandworm and UAC-0099 jointly executed a destructive wiper attack against a Ukrainian logistics company using a novel mesh network backdoor — the operation demonstrates continued Russian investment in destructive tooling targeting supply chain sectors.

  • A flaw in the x402 SDK allowed malicious APIs to drain AI agents’ full token balances due to absent spending limits in the default client — organizations deploying autonomous AI payment workflows face direct financial exposure.



Immediate Action Required

  • ConnectWise users: Treat this as a potential active compromise scenario. Audit privileged access, review ConnectWise logs for anomalous activity, and validate whether any managed environments show signs of lateral movement. MSPs must notify downstream customers.

  • VMware VeloCloud Orchestrator operators: Apply available patches immediately. This is the third exploited zero-day in four months — if no patch is available, isolate the orchestrator from internet exposure or restrict access to trusted management networks.

  • PHP web application teams: Confirm PHP2Shell emergency patches are fully deployed. Audit PHP configurations for exposure to CVE-2026-115560 and monitor for vendor patch release. Increase monitoring on internet-facing PHP applications in the interim.

  • Python development and platform teams: Update Pillow to the patched version immediately. Identify all internal services processing user-supplied images — particularly EPS files — and confirm Ghostscript integration is not exploitable in your environment.

  • AI/automation teams using x402 SDK: Confirm you are running the August 18 or later release that introduced spending limits. Audit all autonomous AI agent payment workflows for missing financial controls and enforce per-transaction spending caps.



High-Impact Developments

China APT Weaponizes AI Pentest Tool Artex Against ConnectWise at Scale

  • What happened: A Chinese state-sponsored APT leveraged Artex, an AI-assisted penetration testing tool, to attack ConnectWise — a widely deployed IT management and remote access platform. The campaign targeted hundreds of organizations and resulted in confirmed privileged access across a subset of environments.

  • Why it matters: ConnectWise is foundational infrastructure for managed service providers. Privileged access to ConnectWise translates directly into privileged access to every customer environment managed through it. The use of an AI-assisted pentest tool signals adversary capability maturation and the potential for faster, more adaptive exploitation at scale.

  • Who should care: CISOs and security directors at organizations using ConnectWise directly or through an MSP; SOC leaders assessing whether their environment was accessed via a compromised MSP; managed services leadership.

  • Recommended action: Immediately audit ConnectWise access logs for anomalous privileged sessions. Rotate credentials associated with ConnectWise administrative accounts. MSPs must proactively notify customers and assess whether managed endpoints show signs of unauthorized access. Engage ConnectWise support for available indicators.

  • Confidence: High — active exploitation confirmed per source reporting.

  • Search metadata: ConnectWise, Artex, China-APT, Privilege Escalation

Intelligence Context



VMware VeloCloud Orchestrator: Third Exploited Zero-Day in Four Months

  • What happened: Attackers are actively exploiting a critical SQL injection zero-day in VMware VeloCloud Orchestrator that allows unauthenticated access to privileged internal functionality and full operator-level privilege escalation. This is the third exploited zero-day in this product within a four-month window.

  • Why it matters: VeloCloud Orchestrator is a network management control plane. Operator-level compromise means attackers can manipulate SD-WAN configurations, intercept traffic, and pivot into managed network segments. Three zero-days in four months indicates this product is under sustained, targeted research by threat actors — the attack surface is not shrinking.

  • Who should care: Network operations and infrastructure teams running VeloCloud Orchestrator; SOC leaders responsible for network management plane monitoring; security architects evaluating SD-WAN risk posture.

  • Recommended action: Apply vendor patches immediately upon availability. If no patch is available, restrict Orchestrator management access to trusted IP ranges and disable internet-facing exposure where operationally feasible. Review logs for unexplained Orchestrator activity over the past four months and treat anomalies as potentially malicious.

  • Confidence: High — active exploitation confirmed.

  • Search metadata: VeloCloud Orchestrator, VMware, SQL Injection, Privilege Escalation, T1190, zero-day

Intelligence Context



PHP Discloses Second Unpatched RCE in 48 Hours — CVE-2026-115560

  • What happened: Two days after shipping emergency patches for the critical PHP2Shell flaws, PHP disclosed CVE-2026-115560, a new unpatched RCE vulnerability. PHP characterizes the risk as low due to its dependency on unusual application configurations, but no patch is yet available.

  • Why it matters: Rapid successive critical PHP RCE disclosures elevate attacker interest in PHP-based web applications broadly. Even with a narrow configuration dependency, CVE-2026-115560’s unpatched state creates a window of elevated risk — particularly for teams running non-standard or legacy PHP configurations that have not been audited.

  • Who should care: Application security teams, web operations, and infrastructure teams running PHP-based applications — particularly those with non-standard or legacy configurations.

  • Recommended action: Confirm PHP2Shell emergency patches are fully deployed. Audit PHP application configurations to identify any matching the unusual profile described for CVE-2026-115560. Monitor for patch release and prioritize deployment. Increase monitoring on internet-facing PHP applications in the interim.

  • Confidence: High — vulnerability confirmed; exploitation status unknown.

  • Search metadata: PHP, CVE-2026-115560, RCE, T1190, Remote Code Execution

Intelligence Context



Critical RCE in Python Pillow Library via Crafted EPS Images — CVE-2026-112900

  • What happened: CVE-2026-112900 in the Python Pillow image library allows remote code execution when a crafted EPS image is processed. The flaw stems from Pillow passing filenames to Ghostscript without adequate sanitization, enabling OS command execution on the server.

  • Why it matters: Pillow is the dominant image processing library in the Python ecosystem. Any Python application that accepts user-supplied images — including EPS files — and processes them server-side is potentially vulnerable. The attack surface is broad given Pillow’s ubiquity across web applications, data pipelines, and document processing services.

  • Who should care: Application security, software engineering, and platform engineering teams running Python applications that handle image uploads or document processing.

  • Recommended action: Update Pillow to the patched version immediately. Identify all services processing user-supplied images, with particular attention to EPS file handling. If EPS processing via Ghostscript is not operationally required, disable it as an interim mitigation.

  • Confidence: High — vulnerability confirmed; exploitation status unknown.

  • Search metadata: Pillow, CVE-2026-112900, Python, RCE, T1190, Ghostscript

Intelligence Context



x402 SDK Flaw Exposes AI Agent Payment Workflows to Full Token Drain

  • What happened: A vulnerability in the x402 SDK — used by AI agents to make API payments — allowed malicious APIs to request and receive an agent’s full token balance. The root cause was the absence of spending limits in the SDK’s default client configuration. A fix was released on August 18 adding spending limits.

  • Why it matters: As organizations deploy autonomous AI agents with payment capabilities, the financial controls governing those agents become a direct security concern. This flaw demonstrates that without explicit spending limits, a compromised or malicious API endpoint can fully drain an agent’s allocated funds. AI agent financial controls are an emerging attack surface that most security programs have not formally addressed.

  • Who should care: AI security leads, application security teams, engineering teams building autonomous AI workflows, and finance teams with visibility into AI operational budgets.

  • Recommended action: Confirm x402 SDK is updated to the August 18 or later release. Audit all AI agent deployments for spending limit configurations. Establish policy requiring explicit per-transaction and per-session spending caps on all autonomous AI payment workflows. Review API integrations used by AI agents for trustworthiness.

  • Confidence: High — vulnerability confirmed by security research; exploitation status unknown.

  • Search metadata: x402 SDK, AI-agents, Financial Theft, API Abuse, token-theft

Intelligence Context



Monitor Only

  • Sandworm and UAC-0099 jointly executed a destructive wiper attack against a Ukrainian logistics company in July 2025 using a novel mesh network backdoor — logistics and critical supply chain organizations should track this TTP development for geopolitical spillover risk. Source: Russian-Linked Hackers Used New Mesh Network Backdoor in Destructive Ukraine Attack — The Hacker News — https://thehackernews.com/2026/10/russian-linked-hackers-used-new-mesh.html



Analyst Observation

Today’s brief reflects a threat environment where management and automation platforms — ConnectWise, VeloCloud Orchestrator, x402 SDK — are the primary targets, not end-user systems. Attackers are investing in tools like Artex that accelerate exploitation at scale, and VeloCloud’s third zero-day in four months is a clear signal that some products are under active, sustained research by capable adversaries. The PHP and Pillow RCE disclosures arriving in the same window are not coincidental noise — they reflect the compounding risk that emerges when multiple critical flaws surface in rapid succession across widely deployed open-source and scripting infrastructure. Security leaders should resist treating these as isolated patch tickets and instead use this moment to pressure-test whether their vulnerability management programs can close critical gaps in web and Python infrastructure within days, not weeks.





Generated by DevSecOpsDadAttack cyber threat intelligence.

Tags:  Cyber-Security-News  CVE-2026-115560  CVE-2026-112900  T1190  VMware  Linux  PHP  RCE  Remote-Code-Execution  Sandworm  UAC-0099  Ukraine
DevSecOpsDadAttack!
Share: X (Twitter) LinkedIn