The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added critical vulnerabilities affecting Trend Micro Apex One and the open-source AI orchestration platform Langflow to its Known Exploited Vulnerabilities (KEV) catalog, confirming that threat actors are actively exploiting flaws in both enterprise endpoint defenses and emerging AI development pipelines.
The update includes CVE-2025-34291, a vulnerability in Trend Micro Apex One with a CVSS severity score of 9.4. Alongside these entries, CISA also added flaws affecting Windows Shell and ConnectWise ScreenConnect to the catalog, reinforcing its role as a curated registry of weaknesses with verified real-world exploitation rather than theoretical risk.
For security operations teams, a KEV listing is a clear signal to initiate out-of-cycle emergency patching. The catalog exists precisely to cut through vulnerability noise and highlight threats that demand immediate action. Organizations running Apex One should cross-reference asset inventories against the updated catalog and deploy patches outside standard maintenance windows. Forensic reviews of affected systems are also warranted to verify no prior compromise occurred before remediation.
The Langflow inclusion marks a notable expansion in the types of software CISA treats as critical. Langflow is an open-source framework for building and orchestrating AI workflows, and its presence in the KEV catalog confirms that adversaries now view AI development environments as direct extensions of traditional attack surfaces. Unlike commercial vendors with predictable patch cycles, open-source AI frameworks rely on community-driven remediation, creating a structural gap that leaves production deployments exposed.
Until official patches are available, organizations using Langflow must deploy compensating controls: strict network segmentation, authenticated API access restrictions, and continuous dependency scanning. Security programs should formally integrate AI supply chain monitoring into existing vulnerability management workflows, treating all KEV catalog additions as critical operational emergencies.
This dual listing — spanning a mature enterprise security product and a rapidly adopted AI framework — reflects how threat actors no longer distinguish between legacy infrastructure and emerging technology. They are exploiting both in parallel, and security teams must respond accordingly. Emergency patching for CVE-2025-34291 should take immediate precedence, while AI framework deployments require formal governance protocols that include rapid response playbooks and ongoing dependency oversight.
美國網絡安全及基礎設施安全局（CISA）已將影響Trend Micro Apex One及開源AI編排平台Langflow的關鍵漏洞納入其已知被利用漏洞（KEV）目錄，確認威脅行為者正積極利用企業端點防護及新興AI開發pipeline中的缺陷。
是次更新包括CVE-2025-34291，一個存在於Trend Micro Apex One的漏洞，CVSS嚴重程度評分為9.4。CISA同時將影響Windows Shell及ConnectWise ScreenConnect的漏洞加入目錄，進一步確立該目錄作為經過篩選的漏洞登記冊，專門記錄經驗證在現實世界中被利用的弱點，而非理論風險。
對security operations團隊而言，KEV列表是啟動非週期性緊急patching的明確信號。該目錄的存在正是為了過濾漏洞噪音，突顯需要立即行動的威脅。運行Apex One的機構應將資產庫存與更新後的目錄進行交叉比對，並於標準維護窗口外部署patches。同時亦應對受影響系統進行forensic審查，以核實在修復前未有遭受入侵。
納入Langflow標誌著CISA視為關鍵的軟件類型出現顯著擴展。Langflow是一個用於構建和編排AI workflows的開源framework，其出現在KEV目錄中確認對手現已將AI開發環境視為傳統attack surfaces的直接延伸。與具有可預測patching週期的商業供應商不同，開源AI框架依賴社群驅動的修復，造成結構性缺口，令production deployments暴露於風險之中。
在官方patches推出前，使用Langflow的機構必須部署compensating controls：嚴格的network segmentation、經認證的API存取限制，以及持續的dependency scanning。Security programs應正式將AI supply chain monitoring整合至現有的vulnerability management workflows，將所有KEV目錄新增項目視為關鍵operational emergencies處理。
此雙重列表——涵蓋成熟的企業security product及迅速被採用的AI framework——反映threat actors已不再區分legacy infrastructure與新興技術。他們正同時利用兩者，security teams必須相應作出回應。CVE-2025-34291的緊急patching應即時優先處理，而AI framework deployments則需要建立正式的governance protocols，包括rapid response playbooks及持續的dependency oversight。