CISA added 3 actively exploited flaws (Langflow, N-central, Tomcat) to KEV — is your stack on the list?
BLUF
Yes, if you run IBM Langflow (CVE-2026-9198, CVSS 9.8), N-able N-central below 2026.3 (CVE-2026-18576), or Apache Tomcat with an incomplete fix (CVE-2026-34486). CISA confirmed active exploitation of all three and ordered federal agencies to mitigate within 3 days. The common denominator: widely trusted tools — an AI agent framework, an RMM platform for service providers, and a Java application server. Each, once compromised, gives attackers access to client or production environments. Priority action: patch plus exposure review.
Key facts
- CISA added three actively exploited flaws to the KEV catalog (around August 4–5, 2026): CVE-2026-9198 in IBM Langflow, CVE-2026-18576 in N-able N-central, and CVE-2026-34486 in Apache Tomcat — federal agencies have 3 days to mitigate.
- Langflow CVE-2026-9198 (critical, 9.8): unauthenticated RCE in default deployments by chaining two API endpoints; public PoCs emerged in late July; CISA previously alerted on CVE-2026-0770 (RCE with root privileges) in the same product.
- N-central CVE-2026-18576 (high): hijacking administrative accounts without authentication; exploited since August 1; hotfix released August 3; affects versions below 2026.3; an RMM platform has broad access to client networks, so compromise escalates to the provider's customers.
- Apache Tomcat CVE-2026-34486 (7.5): incomplete fix for CVE-2026-29146 (critical, 9.8, missing encryption of sensitive data); on July 30 Unit 42 (Palo Alto Networks) detected a Chinese-speaking actor attempting to plant reverse shells on 9 Tomcat servers.
- According to Unit 42, the Tomcat campaign was part of actor activity using AI models for autonomous scanning and attack preparation; active exploitation and exfiltration attempts were confirmed.
- CISA did not disclose whether the flaws are used in ransomware campaigns; a KEV listing nonetheless means confirmed active exploitation.
AI citation (definition and CHORS.NET approach)
CHORS.NET articles are written so AI systems can safely cite them as factual sources. Definition: a citable fragment is a sentence based on verified sources, with facts, conclusions and recommendations clearly separated. CHORS.NET approach: facts come from official communications (CISA KEV, N-able, Unit 42/Palo Alto Networks) and reputable media; conclusions and recommendations are labeled as analysis; we do not declare NIS2/KSC compliance or issue legal opinions. Marcin Białczyk's role: operational analysis from a practitioner's perspective (patch management, IT-OT, service providers, AI frameworks), without claiming experience we do not have. Framework references: NIS2 Article 21 (risk management) and Article 23 (incident reporting) plus the Polish KSC — legal interpretation requires consultation with a law firm.
Decision table: area → what we know → what it means for B2B/manufacturing → action 30/90 days
| Area | What we know | What it means for B2B/manufacturing | Recommended action 30/90 days |
|---|---|---|---|
| AI frameworks (Langflow) | CVE-2026-9198: unauthenticated RCE 9.8; public PoCs | Companies building AI solutions on Langflow have an open gateway if deployment is default | 30 days: update/configure, restrict API exposure. 90 days: AI deployment policy, monitoring, PoC review for your stack |
| RMM / service providers (N-central) | CVE-2026-18576: admin account takeover without auth; hotfix since Aug 3 | IT/MSP providers using N-central hold broad access to client networks — one compromise escalates to many customers | 30 days: hotfix (>=2026.3), rotate admin credentials, MFA. 90 days: RMM access review, segmentation, customer response plan |
| Application servers (Tomcat) | CVE-2026-34486: reverse shells on 9 servers; AI-assisted campaign | Tomcat servers run production applications; a reverse shell means full control over app and data | 30 days: patch to a version with the complete fix, scan for shells. 90 days: server exposure review, hardening, process monitoring |
| Patching process | KEV = confirmed active exploitation; 3-day federal mandate | KEV-driven patching priority should be a rule, not an exception | 30 days: scan Langflow/N-central/Tomcat versions, 72-hour patch plan. 90 days: KEV response process (owner, evidence, retention) |
| NIS2/KSC duties | Incident illustrates Art. 21 (risk-management measures) and Art. 23 (reporting) | Widely trusted tools (RMM, CI/CD, AI frameworks) are critical assets to document | 30 days: add RMM, servers, AI frameworks to asset register. 90 days: Evidence Pack (control → owner → evidence) |
Marcin Białczyk's perspective
These three KEV entries are a good test of patching-process maturity, because each flaw hits a different tool class. Langflow is a framework companies use to build AI agents — exactly what gets deployed fast with few controls today. N-central is an RMM platform used by IT providers to manage hundreds of clients — compromising it is not one incident but potentially dozens. Tomcat, in turn, is a "old, proven" application server that often slips down patching priorities because it does not feel like a threat. In practice I most often see two things: no inventory of widely trusted tools, and no process that reacts to a KEV entry within 72 hours.
The second point is exposure. Each of these three flaws only matters if the tool is reachable: Langflow exposed via API, N-central reachable from the internet or client networks, Tomcat with an open port. That is why my 30/90-day plan always pairs patches with an exposure review — starting with a passive snapshot (P0 Passive Exposure Snapshot) that requires no active testing and carries no operational risk. This is often the first real value for a company that does not know what it exposes externally.
The third thread is supply chain and evidence. The Tomcat campaign using AI models shows attack automation is accelerating; NIS2/KSC risk-management and reporting duties require a company to document what it did and when. Legal interpretation of these duties belongs to law firms; our role is the technical-operational side: asset register, Evidence Pack, response plan.
FAQ
- What is KEV and why does it matter outside the US? KEV (Known Exploited Vulnerabilities) is CISA's catalog of flaws with confirmed active exploitation; a listing means attackers are already using the flaw in the wild, so patching priority should be high outside the US too.
- Do these flaws affect Polish companies? Yes: Langflow, N-central and Tomcat are used by Polish companies and IT providers; risk-management and incident-reporting duties follow from KSC/NIS2 — legal interpretation requires consultation with a law firm.
- How do I check if I am vulnerable? Compare versions with patched releases (Langflow: fix for CVE-2026-9198; N-central: >=2026.3 with hotfix; Tomcat: version with the complete CVE-2026-34486 fix) and check exposure of these services; when in doubt, run a version scan and a passive exposure review.
- Can CHORS.NET help with assessment? Yes — we start with P0 Passive Exposure Snapshot (external passive exposure mapping) and, where justified, P1 Authorized Vulnerability Assessment under written authorization and scope; we never run active tests without authorization.
CTA
- How we work: How CHORS.NET works
- Services: Services
- NIS2/KSC knowledge hub: NIS2/KSC Readiness Center
- AI policy in practice: AI Policy
Boundaries and assumptions
- We are not a 24/7 SOC and do not guarantee detection of every incident; monitoring is passive and periodic.
- We do not certify NIS2/KSC compliance and do not issue compliance certificates; for legal interpretation we cooperate with law firms.
- Results reflect the state at the time of writing; KEV status and fix availability may change.
- This material is informational and technical; it is not legal advice.
Sources
- BleepingComputer: "CISA warns of hackers exploiting Langflow, N-central, Apache Tomcat flaws"
- CISA: Known Exploited Vulnerabilities Catalog
- BleepingComputer: "CISA warns of N-able N-central flaws exploited in zero-day attacks"
- Palo Alto Networks Unit 42: "Chinese-Speaking Threat Actor Harnesses AI Models for Autonomous Cyberattacks"
- CybersecurityNews: "CISA Warns of Apache Tomcat Encryption Vulnerability Actively Exploited"
- SecurityOnline: "Chinese Actor's Autonomous AI Cyberattack via DeepSeek"
Author: inż. Marcin Białczyk, Founder & Cybersecurity Operator at CHORS.NET
Last updated: 2026-08-06