Loki Intelligence — Security Briefs · Published
Daily Security Brief: Linux Kernel KEVs, Workflow Engine RCE, and AI Agent Plugins
Today's picture combines actively exploited Linux kernel and workflow-engine flaws with critical cloud AI fixes and a plugin-integrity weakness in AI coding agents. Teams should patch KEV-listed kernels and Orkes Conductor first, update SolarWinds ARM, and review how developer tooling verifies third-party plugins.
Signal 01 · The Hacker News
Microsoft Patches CVSS 10.0 Azure AI Foundry Flaw Enabling Unauthorized Privilege Escalation
Microsoft fixed a maximum-severity missing-authentication flaw in Azure AI Foundry (CVE-2026-85889, CVSS 10.0) that could let an unauthorized attacker escalate privileges over a network. The service-side fix needs no customer action and no exploitation is known. Separate out-of-band Windows 11 26H1 updates address two local privilege escalation bugs.
Why it matters: Cloud-side fixes need no patching, but teams should confirm Windows 11 26H1 devices received the out-of-band update and review who holds access to AI Foundry resources.
Source: The Hacker News
Signal 02 · The Hacker News
Can You Prove a New CVE Is Exploitable Before Attackers Do? Learn How in This Webinar
A partner-contributed webinar announcement argues that a severity score alone does not show whether a new CVE is exploitable in a specific environment. The session, led by a Picus solutions architect, covers mapping a vulnerability to its attack techniques and testing whether existing controls block them when running live exploits is unsafe.
Why it matters: Prioritization improves when exposure and control effectiveness are validated alongside CVSS scores, so remediation effort follows real exploitability in your own environment.
Source: The Hacker News
Signal 03 · CISA
CISA Adds One Known Exploited Vulnerability to Catalog
CISA added CVE-2025-39682, a Linux kernel improper check for unusual or exceptional conditions, to its Known Exploited Vulnerabilities catalog based on evidence of active exploitation. CISA encourages all organizations, not only federal agencies, to build KEV entries into risk-based vulnerability management and remediate them first.
Why it matters: Check Linux servers, cloud images and appliances for vendor kernel updates covering this CVE, and prioritize hosts where untrusted users or workloads can run code.
Source: CISA
Signal 04 · CISA
CISA Adds Two Known Exploited Vulnerabilities to Catalog
CISA added two Linux kernel vulnerabilities to its KEV catalog after confirming active exploitation: CVE-2025-39964, a race condition, and CVE-2026-53266, an out-of-bounds write. The alert repeats CISA's call for every organization to prioritize remediation of catalog entries on exposed and high-value systems.
Why it matters: Confirm distribution kernel packages that fix both CVEs are installed and the hosts rebooted, starting with shared, multi-tenant and container infrastructure.
Source: CISA
Signal 05 · The Hacker News
SolarWinds Patches ARM Hard-Coded Key Flaw Enabling Unauthenticated RCE
SolarWinds patched CVE-2026-28326 (CVSS 8.8), a hard-coded static key in Access Rights Manager that could enable unauthenticated remote code execution. ARM 2026.2 and earlier are affected and the fix ships in 2026.2.1; SolarWinds has not reported exploitation. Recent updates also fixed Web Help Desk and Serv-U flaws.
Why it matters: ARM holds sensitive permission data, so upgrade to 2026.2.1, restrict network access to the server, and confirm Web Help Desk and Serv-U run fixed builds.
Source: The Hacker News
Signal 06 · The Hacker News
Critical Pre-Auth RCE in Orkes Conductor Workflow Platform Exploited in the Wild
Fortinet reports active exploitation of CVE-2026-58138 (CVSS 9.8), an unauthenticated remote code execution flaw in Orkes Conductor versions from 3.21.21 before 3.30.2. Attackers submit crafted workflow definitions to the workflow API, abusing unsandboxed script evaluators to run OS commands; Fortinet blocked nearly 7,000 attempts in one week.
Why it matters: Upgrade to Conductor 3.30.2 or later; until then, keep the workflow API off untrusted networks and monitor for unexpected workflow submissions and command execution.
Source: The Hacker News
Signal 07 · The Hacker News
Plugin4Shell Lets Repository Owners Swap Pinned Plugin Code Across Four AI Coding Agents
Air Security says four AI coding agents install plugins pinned to a commit hash without checking that the fetched code matches, letting a repository owner on Git hosts that allow hash-like branch names swap the plugin. Per Air, Claude Code 2.1.179 and Codex 0.146.0 fix this; Copilot is unpatched and Gemini CLI will not be fixed.
Why it matters: Update affected agents, prefer plugins from GitHub-hosted marketplaces, and treat plugins as code running with developer credentials that deserves review and least privilege.
Source: The Hacker News
Brief sources
Related briefs
- What Is AI Agent Security Testing?
- AI Red Teaming vs Traditional Pentesting
- MCP Security Testing Checklist
Relevant Loki service: Orvyn — AI agent security private preview.