Weekly Threat Bulletin – February 18th, 2026

CISA Flags Critical Microsoft SCCM Flaw as Exploited in Attacks

CISA has mandated U.S. federal agencies to secure their systems against a critical Microsoft Configuration Manager vulnerability, CVE-2024-43468, which is now actively exploited in attacks. This SQL injection flaw, reported by Synacktiv and patched by Microsoft in October 2024, allows unauthenticated remote attackers to achieve code execution with the highest privileges on the server or underlying database. Despite Microsoft's initial assessment that exploitation was "less likely," Synacktiv released proof-of-concept code in November 2024, leading CISA to flag the vulnerability as actively exploited and order Federal Civilian Executive Branch agencies to patch by March 5th, as per Binding Operational Directive (BOD) 22-01, while also urging all network defenders, including those in the private sector, to apply mitigations promptly.

Severity: Critical

Threat Details and IOCs

Mitigation Advice

• Immediately apply the security updates for CVE-2024-43468 to all Microsoft Configuration Manager (SCCM/ConfigMgr) instances as released by Microsoft in October 2024.
• Use a vulnerability scanner to perform an authenticated scan of the network to identify all instances of Microsoft Configuration Manager and confirm they are patched against CVE-2024-43468.
• Audit the underlying SQL database for the Microsoft Configuration Manager site for any unauthorized or recently created sysadmin accounts and monitor for suspicious activity, such as the execution of `xp_cmdshell`.

Compliance Best Practices

• Implement network segmentation to restrict all access to Microsoft Configuration Manager servers and their underlying databases, ensuring they are only accessible from a limited set of authorized administrative workstations via a secure management VLAN.
• Review and update the organization's patch management policy to enforce a shorter deployment timeline for 'critical' and 'exploited' vulnerabilities, especially for management-plane systems like Microsoft Configuration Manager.
• Initiate a recurring audit of service accounts and administrative permissions for all critical infrastructure, including Microsoft Configuration Manager, to ensure they operate with the minimum necessary privileges to perform their functions.
• Enhance the IT asset management program to maintain a real-time, accurate inventory of all software and systems, including versioning, to accelerate identification of vulnerable assets during future security incidents.

Notepad++ - CVE-2025-15556

A critical vulnerability, CVE-2025-15556, exists in Notepad++ versions prior to 8.8.9, stemming from a weakness in the integrity and authenticity verification of updates (CWE-494). This flaw allows an attacker, through interception or redirection of network traffic, to execute an installer under their control, leading to arbitrary code execution with user privileges. The vulnerability carries a CVSS v3.1 score of 8.8, requires no special privileges for exploitation, and has low attack complexity, though user interaction is necessary. It is actively exploited, with evidence suggesting state-linked groups like Lotus Blossom (Billbug) have leveraged it to deploy backdoors. To mitigate this risk, users should update Notepad++ to version 8.8.9 or higher (with some recommendations for 8.9.1 or later to enforce cryptographic signature verification). As a temporary workaround, network access for the `gup.exe` process and/or the `notepad-plus-plus.org` domain can be blocked. This CVE is listed in CISA's Known Exploited Vulnerabilities Catalog, underscoring its severity and the urgency of applying mitigations.

Severity: Critical

Threat Details and IOCs

Mitigation Advice

• Update all instances of Notepad++ to version 8.8.9 or a more recent version.
• Use an Endpoint Detection and Response (EDR) or host-based firewall to create a rule that blocks all outbound network connections from the process `gup.exe`.
• Configure the corporate firewall or DNS filtering service to block all traffic to the domain `notepad-plus-plus.org`.
• Hunt for historical network connections from the process `gup.exe` in endpoint and network logs to identify potential compromises.

Compliance Best Practices

• Implement or enhance a comprehensive software asset inventory system to maintain a real-time catalog of all installed applications and their versions across all company endpoints.
• Establish a formal patch management policy and program that prioritizes vulnerabilities based on active exploitation status and ensures timely deployment of security updates for all third-party software.
• Implement an application allowlisting policy to prevent the execution of unauthorized or unsigned executables and installers on user workstations.
• Develop and enforce a network egress filtering policy that restricts outbound internet access from workstations to only what is required for business functions.
• Enforce the principle of least privilege by ensuring standard users do not have local administrator rights on their workstations.

WordPress Plugin with 900k Installs Vulnerable to Critical RCE Flaw

A critical remote code execution (RCE) vulnerability, tracked as CVE-2026-1357 with a severity score of 9.8, has been identified in the WPvivid Backup & Migration plugin for WordPress, affecting over 900,000 installations. This flaw allows unauthenticated attackers to upload arbitrary files, potentially leading to a complete website takeover, particularly on sites with the non-default "receive backup from another site" option enabled. The vulnerability stems from improper error handling in RSA decryption, which results in a predictable encryption key, combined with a lack of path sanitization that permits directory traversal and the upload of malicious PHP files. Although exploitation is limited by a 24-hour key validity window, the feature is commonly used for site migrations. The issue was reported by Lucas Montes (NiRoX) to Defiant on January 12, and the vendor, WPVividPlugins, released a fix in version 0.9.124 on January 28, which includes checks for RSA decryption failures, filename sanitization, and restrictions on uploaded file types. Users are strongly advised to upgrade to version 0.9.124 immediately to mitigate this risk.

Severity: Critical

Threat Details and IOCs

Mitigation Advice

• Identify all WordPress sites using the 'WPvivid Backup & Migration' plugin and update them to version 0.9.124 or newer immediately.
• Review all instances of the 'WPvivid Backup & Migration' plugin and disable the 'receive backup from another site' setting unless it is actively required for a scheduled migration.
• Scan web server file systems for any recently created or modified PHP files within WordPress directories that do not correspond to a legitimate plugin or core update.

Compliance Best Practices

• Establish a formal vulnerability management program that includes automated scanning and a defined patching schedule for all third-party software, including WordPress plugins and themes.
• Deploy a Web Application Firewall (WAF) and create rules to block or alert on directory traversal attempts and the uploading of executable file types like PHP to non-standard directories.
• Implement a File Integrity Monitoring (FIM) solution on all web servers to alert on the creation of or changes to files within the web application's directories.
• Implement a security policy that mandates periodic review of all application and plugin configurations to ensure non-essential features are disabled, following the principle of least privilege.

Inside the ‘graphalgo’ Fake Crypto Developer Recruitment Campaign

The 'graphalgo' campaign, attributed to the North Korean Lazarus Group, is an ongoing fake cryptocurrency developer recruitment operation targeting JavaScript and Python developers through social platforms like LinkedIn and Facebook, and job forums such as Reddit. This campaign employs a fabricated narrative around a blockchain and cryptocurrency exchange company, distributing a Remote Access Trojan (RAT) via 192 malicious packages across npm and PyPI. The campaign began on May 2, 2025, on npm with the `graphalgo` package, later expanding to include packages with "big" in their names starting November 2025. Initial packages, like `graphalgo`, mimicked legitimate libraries and executed a malicious payload (`graph-init.min.js`) upon object instantiation, downloading a non-malicious artifact from GitHub to derive a second-stage payload's IP address. Persistence is established by saving scripts to specific paths on Windows, Linux, and Mac systems, such as `<homeDir>/AppData/Local/Google/Chrome/User Data/Scripts/startup.js`. Later packages, exemplified by `graphnetworkx`, introduced an encryption layer, requiring a specific decryption key constructed from arguments passed to the graph constructor (e.g., `weighted-directed-graph`). A second-stage payload is downloaded from ``hxxps[:]//raw[.]githubusercontent[.]com/johns92/blog_app/refs/heads/main/server/.env.example`,` leading to the final RAT payload that periodically fetches and executes commands from a command and control server. The attackers heavily utilized GitHub to host infrastructure, including fake hiring tasks that provided the necessary arguments to activate the malware, and each malicious package had a corresponding GitHub repository. Malicious scripts are designed to delete themselves and patch invoking files to evade detection. PyPI variants share similar naming conventions, malicious payloads, code flow, key construction, C2 infrastructure, and payload removal functions with their npm counterparts, though decryption functions differ.

Severity: Critical

Threat Details and IOCs

Mitigation Advice

• Add the URL `raw.githubusercontent.com/johns92/blog_app/refs/heads/main/server/.env.example` to the network firewall and web proxy blocklists.
• Scan developer workstations and build servers for the existence of `startup.js` files in the specific persistence locations identified for Windows, Linux, and macOS.
• Audit all JavaScript project dependencies (`package.json` files) to identify and remove any packages containing 'graph' or 'big' in their names, paying special attention to those explicitly named in the report.
• Audit all Python project dependencies (`requirements.txt` or similar files) to identify and remove any packages containing 'graph' or 'big' in their names.

Compliance Best Practices

• Implement a software supply chain security tool to automatically scan open-source packages from repositories like npm and PyPI for malicious code, suspicious behaviors, and vulnerabilities before they are approved for developer use.
• Develop and conduct regular security awareness training for all developers, focusing on identifying social engineering attempts on professional networking sites and the security risks of using unvetted, third-party code.
• Configure and tune Endpoint Detection and Response (EDR) policies to detect and alert on suspicious process execution, such as scripts running from unusual user profile directories or processes making unexpected outbound network connections.
• Establish a formal policy governing the use of open-source software, requiring that all new packages undergo a security review and are sourced from a trusted internal artifact repository rather than directly from public repositories.

42,900 OpenClaw Exposed Control Panels and Why You Should Care

The rapid, unsupervised deployment of AI agents like OpenClaw (formerly Moltbot and Clawdbot) has led to 42,900 exposed control panels across 82 countries, with 78% of these instances remaining unpatched against known vulnerabilities. These agents, designed for ease of use by non-technical staff, often store credentials insecurely (sometimes in plaintext local directories or backup files) and operate with broad permissions, creating a significant "visibility gap" in enterprise security. Traditional security models, built for human-paced sessions and managed endpoints, are ill-equipped to handle these autonomous, continuously operating entities that bypass standard IAM workflows and whose activity appears legitimate in logs. Specific vulnerabilities include CVE-2026-25253 (one-click token theft leading to takeover, even on locally bound instances), CVE-2026-25157 (command injection on macOS developer workstations), and CVE-2026-24763 (container escape risk). When compromised, these agents act as "multipliers," granting attackers access to an entire digital workspace, including email, cloud accounts, internal chat, and browser sessions, and can facilitate lateral movement and social engineering. To mitigate this, organizations must immediately scan for OpenClaw signatures (e.g., port 18789) using tools like Shodan or Censys, update to version 2026.1.29 or later, bind the gateway to 127.0.0.1, set strong authentication, rotate all touched credentials, and purge backup files. Furthermore, it is crucial to establish continuous visibility by treating AI agents as privileged non-human identities, adding agent detection to endpoint monitoring, scanning for credential directories, and updating security architecture to aggressively segment agent access, log agent actions, and revise incident response playbooks for agent compromise scenarios.

Severity: Critical

Threat Details and IOCs

Mitigation Advice

• Use external scanning tools like Shodan or Censys to scan all company IP ranges for indicators of OpenClaw, Moltbot, and Clawdbot, specifically looking for open port 18789 and associated HTML titles or favicon hashes.
• For any discovered instances of OpenClaw, Moltbot, or Clawdbot, immediately update them to version 2026.1.29 or later to patch critical vulnerabilities.
• On any discovered OpenClaw instance, configure the service to bind its gateway to the localhost interface (127.0.0.1) and enforce a strong, unique authentication password.
• For any system where an OpenClaw agent was found, immediately rotate all credentials the agent had access to, including API keys, OAuth tokens, SSH keys, and cloud service credentials.
• Use endpoint detection and response (EDR) or scripting to scan all workstations and servers for the presence of credential directories such as ~/.openclaw/, ~/.clawdbot/, and ~/.moltbot/.
• On any host where OpenClaw was installed, manually search for and securely delete any credential backup files (e.g., .bak files) within the agent's data directories to prevent the recovery of old tokens.

Compliance Best Practices

• Develop and implement a formal corporate policy that governs the evaluation, approval, and secure deployment of all AI agents and automation tools.
• Update your identity and access management (IAM) program to classify AI agents as privileged non-human identities, ensuring they are subject to the same inventory, monitoring, and access governance as service accounts.
• Enhance endpoint monitoring by creating and deploying detection rules that look for behavioral patterns of autonomous agents, such as high-frequency API calls, cross-application automation, or off-hours activity from user workstations.
• Update your incident response playbooks with specific procedures for containing and remediating compromised AI agents, including steps for multi-service credential rotation and identifying the full scope of an agent's permissions.
• Implement network and application-level segmentation to ensure that AI agents operate with the principle of least privilege and cannot access multiple disparate services (e.g., email, cloud storage, internal chat) simultaneously.
• Configure application and cloud service logging to capture detailed actions performed by AI agents, and ingest these logs into your SIEM to build detections based on agent-specific activity rather than just user authentication events.