Weekly Security Intelligence Briefing — Week of 2026-03-02

“`html

Executive Summary

The week of March 2-12, 2026 produced a high-density threat environment demanding immediate attention across network infrastructure, mobile platforms, enterprise software, and the supply chain. The most urgent items are two Cisco Secure Firewall Management Center vulnerabilities carrying a perfect CVSS 10.0 score with no available workaround, and a VMware Aria Operations RCE flaw already confirmed by CISA in the Known Exploited Vulnerabilities (KEV) catalog with a March 24 remediation deadline. March Patch Tuesday delivered 79-84 vulnerabilities across Microsoft and Adobe products, including two actively exploited zero-days. CISA simultaneously added legacy Hikvision camera and Rockwell Automation Logix controller flaws to KEV, both confirmed exploited in OT/ICS environments with a March 26 deadline. The threat intelligence picture is equally concerning: an AI-native attack platform (CyberStrikeAI) has compromised more than 600 FortiGate devices; Iran launched a multi-actor cyber campaign coinciding with military operations; APT28 deployed a custom Covenant C2 variant; and the supply chain saw malicious npm packages, poisoned Chrome extensions, and fake AI coding tool installers targeting developers. Overall risk posture is HIGH, trending toward CRITICAL for organizations running Cisco, VMware, or internet-exposed OT assets.

Critical Action Items

1. Patch Cisco Secure Firewall Management Center immediately, CVSS 10.0, no workaround.
   CVE-2026-20079 and CVE-2026-20131 allow unauthenticated remote code execution with full confidentiality, integrity, and availability impact (CVSS:3.1 AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H). No workaround exists. Apply the Cisco March 2026 Bundle Advisory patch now. Prioritize internet-facing FMC instances. Confirm patch status in your CMDB before end of business today. Source: Cisco PSIRT Advisory (SCC-CVE-2026-0004).
2. Remediate VMware Aria Operations RCE (CVE-2026-22719), CISA KEV deadline March 24.
   This command-injection vulnerability (CVSS 8.1) enables unauthenticated RCE and is confirmed exploited in the wild. Federal agencies are required to patch by March 24; all organizations should treat this deadline as a hard target. Apply Broadcom/VMware’s released patch. If patching is not immediately possible, isolate Aria Operations management interfaces from untrusted networks. Source: CISA KEV Catalog (SCC-CVE-2026-0001).
3. Apply March 2026 Microsoft Patch Tuesday updates, two actively exploited zero-days included.
   The March 2026 release addresses 79-84 vulnerabilities across Windows, Office, and Azure components, including two zero-days confirmed exploited before patches were available. Prioritize the zero-day CVEs (see Section 4 table). Deploy via WSUS/Intune within your standard emergency patch window (recommended: 72 hours for exploited CVEs). Note: Microsoft will enable Windows hotpatch by default beginning May 2026 for Intune-managed devices, begin eligibility review now. Source: Microsoft Security Update Guide, SCC-STY-2026-0001 through 0007.
4. Remediate CISA KEV additions: Hikvision cameras (CVE-2017-7921) and Rockwell Logix (CVE-2021-22681), deadline March 26.
   Both OT/ICS flaws are confirmed actively exploited. CVE-2017-7921 (CVSS 9.8) allows unauthenticated access to Hikvision camera streams and credentials. CVE-2021-22681 allows remote attackers to forge Logix controller authentication. Audit your OT asset inventory, apply vendor patches where available, segment affected devices from IT and internet-accessible networks, and verify no direct internet exposure. Federal agencies must comply by March 26. Source: CISA KEV Catalog (SCC-CVE-2026-0005).
5. Investigate FortiGate device integrity, CyberStrikeAI campaign has compromised 600+ devices.
   An AI-native attack platform has achieved large-scale FortiGate compromise, confirmed as the week’s highest-priority SCC campaign item (priority score 0.92). Review FortiGate firmware versions, check for unauthorized configuration changes, audit admin accounts and API tokens, and verify integrity of management-plane access logs. Cross-reference IOCs from the SCC-CAM-2026-0001 campaign brief. Source: SCC Intelligence Pipeline (SCC-CAM-2026-0001).
6. Audit npm dependencies and browser extensions for supply chain compromise.
   A malicious npm package impersonating an OpenClaw AI installer deployed a RAT and stole macOS credentials. Separately, two Chrome extensions turned malicious after ownership transfer. Run npm audit and review your approved extension list. Remove or block @openclaw-ai/openclawai from all developer environments. Enforce a browser extension allowlist via policy. Review CI/CD pipelines for unauthorized package additions. Sources: The Hacker News (npm package), The Hacker News (Chrome extensions).

Key Security Stories

1. March 2026 Patch Tuesday: 84 Vulnerabilities Including Two Actively Exploited Zero-Days

Microsoft’s March 2026 Patch Tuesday addressed between 79 and 84 vulnerabilities (sources differ slightly in count methodology) across Windows, Office, Azure, and developer tooling. Two of these were confirmed zero-days, exploited in the wild before patches were available. Adobe released companion updates on the same day. The combined release represents one of the larger monthly patch loads in recent cycles and includes remote code execution, privilege escalation, and information disclosure classes across multiple products.

The two actively exploited zero-days are the immediate priority. Cisco Talos published Snort rule coverage for prominent vulnerabilities in the release, which security operations teams should deploy to detection infrastructure immediately alongside patch deployment. Defender for Endpoint and Microsoft Sentinel customers should verify their protection content has been updated to reflect the new CVE set.

Microsoft also announced that Windows hotpatch, which applies security updates without requiring a system reboot, will be enabled by default for all eligible devices managed through Microsoft Intune and the Microsoft Graph API beginning with the May 2026 update cycle. This is a meaningful operational improvement for organizations that have struggled to meet aggressive patch SLAs on server workloads. Organizations should begin evaluating device eligibility now.

• Affected products: Windows (multiple versions), Microsoft Office, Azure components, Adobe Creative Cloud products
• Exploitation status: Two zero-days confirmed exploited in the wild; remainder unconfirmed at time of release
• Fix: Apply March 2026 cumulative updates via Windows Update, WSUS, or Intune
• Sources: SCC-STY-2026-0001 through SCC-STY-2026-0007; Microsoft Security Update Guide; Cisco Talos (Snort rules)

2. CyberStrikeAI: AI-Native Attack Platform Compromises 600+ FortiGate Devices

The highest-priority campaign item this week is CyberStrikeAI, an AI-native attack platform that has confirmed compromises across more than 600 FortiGate devices globally. This represents a qualitative shift in attacker capability: the platform automates reconnaissance, exploitation selection, and payload delivery using AI components, enabling scale and speed that traditional manual campaigns cannot match. The SCC pipeline assigned this a priority score of 0.92, the highest of any item this week.

The campaign targets FortiGate’s management and VPN surfaces. Compromised devices have been observed with unauthorized configuration changes, new administrative accounts, and evidence of credential harvesting. Attribution remains unconfirmed in open sources at time of publication. The AI-driven nature of the platform means victim selection and targeting decisions may be opaque and opportunistic, not limited to specific verticals.

Organizations running FortiGate appliances should treat this as an active threat. Immediate steps include firmware version verification, admin account audit, configuration integrity review, and examination of VPN and management access logs for anomalous activity. Fortinet’s published IOC set should be cross-referenced against SIEM telemetry.

• Affected products: Fortinet FortiGate (multiple versions, confirm against Fortinet PSIRT advisories)
• Exploitation status: Active campaign, 600+ confirmed compromises
• Fix: Apply latest FortiGate firmware; audit accounts and configurations; isolate management interfaces
• Source: SCC Intelligence Pipeline (SCC-CAM-2026-0001)

3. EchoLeak: Zero-Click Microsoft Excel + Copilot Information Disclosure

EchoLeak is a zero-click information disclosure vulnerability affecting Microsoft Excel in combination with Microsoft 365 Copilot. With a CVSS base score of 8.6, the flaw allows an attacker to exfiltrate sensitive data from a victim’s Copilot context, including documents, emails, and AI conversation history, without requiring any user interaction beyond opening a malicious Excel file. The attack surface is particularly concerning given the rapid enterprise adoption of Microsoft 365 Copilot.

The mechanism exploits how Copilot processes embedded prompts within Excel cells or metadata, a class of attack known as indirect prompt injection. This represents the first publicly documented zero-click prompt injection achieving meaningful data exfiltration in a production enterprise AI assistant. Security teams should understand that Copilot’s access to organizational data amplifies the blast radius of any prompt injection flaw significantly beyond what a traditional Office document exploit would achieve.

Microsoft addressed EchoLeak as part of the March 2026 Patch Tuesday release. Organizations using Microsoft 365 Copilot should treat this patch as urgent, ahead of non-exploited CVEs in the same release. Review Copilot licensing and data access scoping to ensure the principle of least privilege is applied to AI assistant data permissions.

• Affected products: Microsoft Excel (Microsoft 365), Microsoft 365 Copilot
• Exploitation status: Not confirmed exploited in the wild at time of publication; zero-click attack chain makes risk elevated
• Fix: Apply March 2026 Microsoft security updates; review Copilot data access scope
• Source: SCC Intelligence Pipeline (SCC-CVE-2026-0007)

4. Android March 2026 Zero-Day: CVE-2026-0006 Actively Exploited (CVSS 9.8)

Google’s March 2026 Android security bulletin included a zero-day vulnerability, CVE-2026-0006, rated CVSS 9.8 (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H) and confirmed actively exploited before patching. The vulnerability allows unauthenticated remote attackers to achieve full compromise of confidentiality, integrity, and availability on affected devices. This is one of the most severe Android CVEs published in recent cycles.

A companion vulnerability, CVE-2026-21385 (Qualcomm Android Kernel Privilege Escalation, CVSS 7.8), was also addressed in the same bulletin. CVE-2026-21385 requires local access and is a local privilege escalation, but in combination with an initial access vector, it extends attacker control to kernel level. Both CVEs are relevant to enterprise mobile device fleets and BYOD environments.

Organizations managing Android devices through MDM should push the March 2026 Android Security Patch Level (SPL) immediately. For BYOD policies, users should be notified and required to demonstrate patch compliance within defined SLA windows. Devices unable to receive the March 2026 SPL should be evaluated for continued enterprise access.

• Affected products: Android (multiple versions); Qualcomm-chipset Android devices (CVE-2026-21385)
• Exploitation status: CVE-2026-0006 confirmed actively exploited; CVE-2026-21385 high-severity, exploitation status unconfirmed
• Fix: Apply Android March 2026 security patch level via OEM update or MDM push
• Sources: SCC Intelligence Pipeline (SCC-CVE-2026-0003, SCC-CVE-2026-0002); Google Android Security Bulletin

5. Ivanti EPM Flaw Confirmed Actively Exploited, CISA Orders Federal Patch

CISA added a high-severity Ivanti Endpoint Manager (EPM) vulnerability to the KEV catalog this week and ordered U.S. federal agencies to patch within three weeks. The flaw joins a broader CISA KEV addition covering SolarWinds and VMware Workspace ONE vulnerabilities (CVE-2021-22054, CVSS 7.5) that were flagged in the same advisory cycle. All three represent confirmed exploitation in the wild, not theoretical risk.

Ivanti has faced a sustained period of critical vulnerability disclosures across its product line. Organizations should review their Ivanti EPM deployment for patch status, audit administrator and agent credentials, and verify that EPM servers are not directly internet-accessible. The SolarWinds and Workspace ONE additions add to an already heavy patching load this week and should be triaged against your specific product inventory.

• Affected products: Ivanti Endpoint Manager; SolarWinds (specific product per CISA advisory); VMware Workspace ONE (CVE-2021-22054)
• Exploitation status: All three confirmed actively exploited per CISA KEV
• Fix: Apply vendor patches per CISA advisory; verify no direct internet exposure for EPM servers
• Sources: CISA KEV Catalog; BleepingComputer (Gatlan, March 10, 2026); The Hacker News (March 10, 2026)

CISA KEV & Critical CVE Table

CVE	Product	CVSS	Status	KEV Deadline	Description
CVE-2026-22719	VMware Aria Operations	8.1	CISA KEV, Actively Exploited	March 24, 2026	Unauthenticated RCE via command injection. Network-accessible. No user interaction required.
CVE-2026-20079	Cisco Secure Firewall Management Center	10.0	Critical, Patch Immediately; No Workaround	N/A (not yet KEV; treat as urgent)	Unauthenticated RCE. Full C/I/A impact. Scope changed. No workaround exists.
CVE-2026-20131	Cisco Secure Firewall Management Center	10.0	Critical, Patch Immediately; No Workaround	N/A (not yet KEV; treat as urgent)	Unauthenticated RCE. Full C/I/A impact. Scope changed. No workaround exists. Bundled with CVE-2026-20079.
CVE-2017-7921	Hikvision IP Cameras	9.8	CISA KEV, Actively Exploited	March 26, 2026	Authentication bypass allows unauthenticated access to camera streams and credentials. Legacy OT/ICS environment risk.
CVE-2021-22681	Rockwell Automation Logix Controllers	9.8 (per NVD/CISA KEV)	CISA KEV, Actively Exploited	March 26, 2026	Forged authentication tokens allow remote attacker to assume trusted device identity on Logix controller. ICS/OT critical risk.
CVE-2026-0006	Android (multiple versions)	9.8	Actively Exploited Zero-Day	N/A (not yet KEV)	Unauthenticated remote exploitation. Full device compromise. Patched in Android March 2026 SPL.
CVE-2026-21385	Qualcomm Android Kernel	7.8	High, Local Privilege Escalation	N/A	Local privilege escalation to kernel level on Qualcomm-chipset Android devices. Post-initial-access risk amplifier.
CVE-2026-20122	Cisco Catalyst SD-WAN Manager	7.1	High, Confirmed Exploited (Active Campaign)	N/A	Confirmed exploited as part of expanding campaign targeting SD-WAN infrastructure. Apply Cisco PSIRT patch.
CVE-2026-20128	Cisco Catalyst SD-WAN Manager	7.1	High, Confirmed Exploited (Active Campaign)	N/A	Second SD-WAN Manager CVE confirmed in same expanding campaign. Bundled remediation with CVE-2026-20122.
CVE-2021-22054	VMware Workspace ONE	7.5	CISA KEV, Actively Exploited	Per CISA directive (verify current KEV listing)	Server-side request forgery (SSRF). Confirmed active exploitation per CISA March 10 advisory.
[March 2026 PT Zero-Day 1]	Microsoft (product per MSRC advisory)	See MSRC	Actively Exploited Zero-Day	N/A (included in Patch Tuesday)	First of two Patch Tuesday zero-days confirmed exploited before patch release. Specific CVE ID not confirmed in SCC pipeline data, verify against Microsoft Security Update Guide.
[March 2026 PT Zero-Day 2]	Microsoft (product per MSRC advisory)	See MSRC	Actively Exploited Zero-Day	N/A (included in Patch Tuesday)	Second Patch Tuesday zero-day confirmed exploited before patch release. Verify CVE IDs against Microsoft Security Update Guide.

Note: Specific CVE IDs for the two Microsoft Patch Tuesday zero-days were not confirmed in SCC pipeline data. Verify current identifiers against the Microsoft Security Update Guide at https://msrc.microsoft.com/update-guide before operationalizing detection rules.

Supply Chain & Developer Tool Threats

Malicious npm Package: @openclaw-ai/openclawai

A malicious npm package named @openclaw-ai/openclawai, uploaded by a user account “openclaw-“, impersonated a legitimate OpenClaw AI installer to deploy a remote access trojan (RAT) and harvest macOS credentials from compromised developer workstations. The package was discovered by threat researchers and reported via The Hacker News. This is a typosquatting and brand impersonation attack targeting developers integrating AI coding assistants, a high-value target given the privileged access developer machines typically hold to source code, secrets, and cloud credentials.

Separately, Microsoft Bing’s AI-enhanced search feature promoted a fake OpenClaw GitHub repository that instructed users to run commands deploying information stealers and proxy malware. The combination of a poisoned search result and a malicious package in npm creates a multi-vector attack surface requiring both endpoint and registry-level controls. Developer machines that executed the install commands or ran code from the promoted repository should be considered compromised until forensically cleared.

Recommended actions: Remove @openclaw-ai/openclawai from all environments. Audit npm install logs for the package name. Review developer workstation endpoint telemetry for RAT indicators. Enforce npm registry restrictions (private registry or package allowlisting) in your CI/CD pipeline. Train developers to verify package publisher identity before installation.

Sources: The Hacker News (npm package, March 9, 2026); BleepingComputer, Toulas (Bing AI / OpenClaw, March 5, 2026)

Chrome Extensions Turn Malicious After Ownership Transfer

Two Google Chrome browser extensions became malicious after their ownership was transferred to a new developer. The new owner pushed an update enabling code injection and data theft capabilities. The extensions were originally associated with a legitimate developer, giving them an established install base and user trust that the new malicious owner exploited as a distribution mechanism. This attack pattern, acquiring extensions with existing user bases and pushing malicious updates, requires no phishing or social engineering of the victim; the browser’s automatic update mechanism delivers the payload.

Recommended actions: Audit your organization’s approved browser extension list. Remove extensions with unclear or recently transferred ownership. Use enterprise browser management (Chrome Enterprise, Edge for Business) to enforce an extension allowlist via policy. Monitor extension update events on managed endpoints. This attack class will likely expand as AI-assisted code review makes malicious update detection harder.

Source: The Hacker News (March 9, 2026)

InstallFix Campaign: Fake Claude Code Sites via Malvertising

The “InstallFix” campaign blends malvertising with a ClickFix-style technique to distribute malware through fake sites impersonating Claude Code, Anthropic’s AI coding assistant. Users searching for Claude Code via web ads are directed to convincing fake sites that instruct them to run a command in their terminal, a ClickFix-style social engineering technique that bypasses browser download warnings entirely because the user executes the payload themselves. This campaign specifically targets developers and technical users who are the most likely to have elevated system access and access to cloud credentials, CI/CD tokens, and source code.

Recommended actions: Brief developer and engineering teams on this specific attack pattern. Enforce DNS filtering to block newly registered domains impersonating AI tooling brands. Disable copy-paste execution from terminal prompts at the policy level where possible. Report suspicious ad placements to your browser vendor.

Source: Dark Reading, Wright (March 9, 2026)

UNC4899 (North Korea): Developer AirDrop Social Engineering Leads to Crypto Firm Breach

North Korean threat actor UNC4899 breached a cryptocurrency organization after a developer at the firm received and installed a trojanized file via AirDrop, Apple’s local wireless file transfer protocol, on their personal device, which was also used for work. The file gave UNC4899 a foothold that expanded into a cloud environment compromise resulting in the theft of millions in cryptocurrency. This represents an evolution in UNC4899’s developer-targeting tradecraft: moving beyond phishing emails to physical-proximity or peer-network social engineering using trusted transfer mechanisms that bypass corporate email and web filters entirely.

Recommended actions: Enforce MDM policies that disable AirDrop on corporate-managed devices. Require physical separation or strict MDM enrollment for developer workstations that access production cloud environments. Review cloud IAM for developer-attributed credential usage from unexpected device types.

Source: The Hacker News (March 9, 2026)

Nation-State & APT Activity Summary

APT28 (Russia / GRU), Custom Covenant C2 Variant for Long-Term Espionage

Russian GRU-linked threat actor APT28 (also known as Fancy Bear) deployed a custom variant of the open-source Covenant post-exploitation framework against targets this week. Covenant is a .NET-based command-and-control framework; APT28’s variant introduces modifications designed to evade signature-based detection tuned to the public version. The campaign is characterized as long-term espionage, APT28 is maintaining persistent access for intelligence collection rather than immediate destructive action. Targeted sectors are not fully confirmed in open sources, but APT28 historically focuses on government, defense, and political organizations in NATO member states.

APT28 also continued a Signal and WhatsApp account hijacking campaign against government officials, military personnel, and journalists, highlighted by a Dutch government warning. The campaign uses phishing techniques to gain access to encrypted messaging account sessions, which contain sensitive communications that are unavailable to standard network interception.

• Attribution: APT28 / Fancy Bear / GRU Unit 26165 (Russia)
• Targeted sectors: Government, defense, political, media (journalists)
• TTPs: Custom Covenant C2 (T1059.001, Command and Scripting Interpreter; T1071, Application Layer Protocol); Signal/WhatsApp linked-device phishing (T1566, Phishing; T1539, Steal Web Session Cookie)
• Detection guidance: Hunt for Covenant-characteristic beacon patterns (HTTP/HTTPS C2 with .NET runtime invocation); monitor for anomalous linked device additions in Signal/WhatsApp administrative logs; alert on messaging app OAuth flows from unexpected IP geolocations

Sources: BleepingComputer, Toulas (APT28/Covenant, March 10, 2026); BleepingComputer, Abrams (Signal/WhatsApp hijacking, March 9, 2026)

Iran Multi-Actor Campaign, Cyber Operations Concurrent with Military Activity

Iranian threat actors launched a coordinated multi-group cyber campaign timed to coincide with military operations. This represents a deliberate integration of cyber and kinetic capabilities. The campaign involves multiple Iranian-affiliated threat actors operating in parallel, suggesting coordination at a strategic level. Historical Iranian cyber operations following military escalation have targeted critical infrastructure, government systems, and media organizations in adversarial states and the region.

• Attribution: Multiple Iranian state-affiliated threat actors (specific group designations not confirmed in SCC pipeline data)
• Targeted sectors: Government, critical infrastructure, media (assess based on geopolitical context)
• TTPs: Coordinated multi-actor operations; likely includes destructive wiper capability based on historical Iranian playbooks (T1485, Data Destruction; T1489, Service Stop); web defacement; DDoS
• Detection guidance: Increase monitoring for wiper-characteristic disk encryption or file overwrite activity; monitor for anomalous outbound data transfers; review perimeter logs for Iranian IP range scanning activity

Source: SCC Intelligence Pipeline (SCC-CAM-2026-0003)

APT36 (Pakistan), AI-Assisted Malware Development (“Vibe-Coding”)

Pakistani threat group APT36 adopted AI-assisted code generation (“vibe-coding”) to produce malware at scale. While the quality of individual malware samples is characterized as mediocre, the volume output enabled by AI tooling is significant enough to overwhelm defensive triage capacity. This is the first publicly reported case of a nation-state APT using AI code generation as a production-scale malware assembly line rather than for targeted development.

• Attribution: APT36 (Pakistan)
• Tactical implication: High-volume, low-quality malware from APT36 requires automated triage and behavioral detection; signature-based approaches will not scale against volume attacks
• Detection guidance: Prioritize behavioral detection rules over signature matching for APT36 infrastructure; enforce application allowlisting to reduce the impact of novel malware variants

Source: Dark Reading, Vijayan (March 5, 2026)

Chinese Threat Actor, Asian Critical Infrastructure Targeting (Multi-Year Campaign)

A Chinese threat actor has conducted a multi-year campaign targeting high-value organizations across South, Southeast, and East Asia. Affected sectors include aviation, energy, government, law enforcement, pharmaceutical, technology, and telecommunications. The campaign uses web server exploits for initial access and deploys Mimikatz for credential harvesting, a relatively well-understood but still highly effective toolchain. The breadth of targeted sectors suggests strategic intelligence collection rather than financially motivated intrusion.

• Attribution: Chinese threat actor (specific group designation not confirmed in SCC pipeline data)
• Targeted sectors: Aviation, energy, government, law enforcement, pharmaceutical, technology, telecommunications
• TTPs: Web server exploits for initial access (T1190); Mimikatz credential dumping (T1003, OS Credential Dumping); long-duration persistence
• Detection guidance: Alert on Mimikatz signatures and LSASS memory access; monitor web servers for exploitation indicators; hunt for living-off-the-land persistence mechanisms (scheduled tasks, WMI subscriptions)

Source: The Hacker News (March 9, 2026)

Phishing & Social Engineering Alert

Microsoft Teams