CISA Orders Urgent Fixes for Exploited Langflow, N-central, and Apache Tomcat Flaws

Federal agencies face tight mitigation deadlines following the addition of three actively exploited security flaws to the Cybersecurity and Infrastructure Security Agency’s Known Exploited Vulnerabilities catalog. The vulnerabilities target critical operational platforms across enterprise IT and modern software stacks: a remote code execution flaw in IBM Langflow (CVE-2026-9198), an administrative account hijacking flaw in N-able N-central (CVE-2026-18576), and an incomplete patch leading to server compromise in Apache Tomcat (CVE-2026-34486).

Threat actors are actively leveraging all three vulnerabilities in live campaigns ranging from manual targeted intrusions by Chinese-speaking threat groups to automated exploitation of publicly exposed AI frameworks. Federal agencies have been ordered to apply available vendor mitigations by Friday, July 7th.

IBM Langflow RCE: AI Framework Under Active Attack

Tracked as CVE-2026-9198, the security flaw in IBM Langflow carries a critical severity rating of 9.8 out of 10. Langflow is a popular visual framework used by developers to orchestrate and build AI agents, process pipelines, and large language model (LLM) integrations. Because these frameworks are designed to handle code execution environments, API keys, and database connectors, securing their underlying runtime is critical.

The flaw allows an unauthenticated, remote attacker to execute arbitrary code on default Langflow deployments. Attackers achieve this by chaining two distinct API endpoints, effectively bypassing authentication controls to run unauthorized commands on the underlying host system. In late July, multiple fully functional proof-of-concept (PoC) exploit scripts emerged publicly, providing step-by-step technical instructions on how to leverage the bypass.

This marks the second critical vulnerability in Langflow targeted in production environments within a short window. CISA issued an alert for CVE-2026-0770, a separate high-impact vulnerability exploited to grant attackers root-level remote code execution. In default architectures, AI orchestration engines often run with elevated host privileges or contain plain-text environment variables holding credentials for vector databases, external SaaS tools, and cloud infrastructure. Compromising the Langflow instance grants adversaries immediate access to pivot deeper into enterprise internal networks.

N-central Account Takeover: Incomplete Patch Exposes RMM Infrastructure

Remote Monitoring and Management (RMM) platforms represent high-value targets for threat actors because they hold centralized, privileged management capabilities across thousands of managed endpoints. The newly listed vulnerability in N-able N-central, tracked as CVE-2026-18576, allows unauthenticated remote attackers to completely hijack administrative accounts.

CVE-2026-18576 represents an incomplete remediation of a previously reported vulnerability. Although N-able initially released a patch to address the underlying account takeover vector, threat actors discovered an alternate bypass technique to circumvent the fix. On August 1st, N-able warned customers of active exploitation in the wild, prompting the vendor to issue an emergency hotfix on Sunday.

The vulnerability impacts all versions of N-central prior to version 2026.3. The blast radius of an RMM administrative account hijack is severe: an adversary who takes over an N-central administrative console gains host-level control over every client system managed by that server. Attackers can push malicious software updates, execute custom scripts, disable endpoint detection and response (EDR) agents, and deploy ransom payloads enterprise-wide without needing to move laterally through standard network protocols.

Apache Tomcat Flaw: Targeted Manual Exploitation by Chinese Threat Actors

The third vulnerability added to the KEV catalog is CVE-2026-34486, a high-severity flaw (CVSS 7.5) affecting Apache Tomcat application servers. Similar to the N-central flaw, CVE-2026-34486 stems from an incomplete patch for an earlier critical issue—CVE-2026-29146 (CVSS 9.8)—which involved missing encryption for sensitive data handled by the server.

On July 30, cybersecurity researchers at Palo Alto Networks Unit 42 reported that a Chinese-speaking threat actor was observed exploiting CVE-2026-34486 in targeted, manual intrusion campaigns. Rather than relying on automated scanning bots, the threat actor manually attempted to exploit nine specific Apache Tomcat web servers to plant interactive reverse shells.

Web server flaws like CVE-2026-34486 are frequently used as initial access vectors by state-sponsored and corporate espionage threat actors. By establishing a reverse shell, the adversary opens an outbound connection from the compromised web server back to actor-controlled command-and-control (C2) infrastructure. This allows the attacker to bypass perimeter firewall ingress rules, conduct local privilege escalation, inspect web application memory for credentials, and maintain long-term persistence within DMZ environments.

Blast Radius and Mitigation Requirements

While CISA has not specified whether these active exploitation campaigns are tied to broader ransomware operations, all three flaws present immediate risks of full network compromise, data exfiltration, and supply-chain escalation. Organizations running affected installations should treat these vulnerabilities as critical emergency items.

To mitigate these risks, security teams should execute the following specific remediation steps immediately:

• IBM Langflow (CVE-2026-9198 & CVE-2026-0770): Identify all internal and publicly exposed instances of Langflow. Restrict network access behind an enterprise VPN or web application firewall (WAF), remove default deployment credentials, and update installations to the latest vendor build addressing both API endpoint chaining and root execution flaws.
• N-able N-central (CVE-2026-18576): Immediately apply the emergency hotfix released by N-able for all N-central versions prior to 2026.3. Ensure that administrative interfaces are strictly restricted to trusted management subnets and verify administrative account audit logs for unauthorized logins or newly created administrative users.
• Apache Tomcat (CVE-2026-34486): Upgrade Tomcat web server instances to the latest release to fix missing encryption and data disclosure paths. Monitor host processes running under the Tomcat user account for unusual child process creation, such as sh, bash, or cmd.exe executing reverse shell commands.