November 19, 2026

From Office NAS Boxes to AI Gateways: CISA Adds Four Actively Exploited Flaws to Its Must-Patch List. Your Questions, Answered

CISA says attackers are already using flaws in Synology DiskStation Manager, Apache Axis, the REDCap research platform and the LiteLLM AI gateway, with ransomware, cryptominers and AI-account hijacking in the mix. We answer what each one is, how it is being used and what to do.

An opened hard disk drive with its silver platter and read arm exposed on a dark workbench

On Wednesday, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) added four vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, the list of flaws it has evidence attackers are actively using. The four cover very different kinds of software, which is exactly why they are worth a closer look. Our answers below are built from CISA’s announcement and the additional exploitation details gathered by The Hacker News.

Which products are on the list?

  • Synology DiskStation Manager (DSM), CVE-2024-10441, the operating system on Synology network storage (NAS) devices. It is an output-encoding flaw rated 9.8 that can let a remote attacker run code.
  • Apache Axis, CVE-2026-90436, an older web services library built into other applications. An unsafe reflection flaw rated 9.8 lets an attacker run code without logging in.
  • REDCap, CVE-2026-90817, Vanderbilt University’s research data capture platform, widely used for surveys and studies. A code injection flaw rated 9.8 allows unauthenticated code execution on the server.
  • BerriAI LiteLLM, CVE-2026-40217, a popular gateway for routing requests to AI models. A flaw rated 8.8 lets a remote attacker run code by rewriting bytecode through the /guardrails/test_custom_code endpoint.

Who is using them, and for what?

Each flaw has its own story. Synology disclosed earlier this month that the DSM bug was being exploited in ransomware attacks on NAS devices, with victims hit by C3RB3R ransomware. The REDCap flaw was used, according to a Safe Security report from late October, to plant a temporary web shell and then a cryptocurrency miner (XMRig) disguised as image files and renamed to look like the syslog-ng logging service, kept alive with a scheduled task and an added SSH key. The LiteLLM bug features in a Google Threat Intelligence Group report this week on “LLMjacking,” in which attackers turned limited AI inference keys into full administrator credentials at financial services, retail, transportation and technology organizations across the Americas and Asia Pacific.

Why does an old library like Apache Axis matter?

Because it hides inside other products. Rapid7 found attackers exploiting the Axis flaw against an Atlassian Confluence Data Center server at a customer, through Confluence’s XML-RPC and SOAP remote API, which uses Axis. Attackers have also reportedly paired it with the “4EverAble2Shell” flaws in N-able N-central to drop cryptominers. Atlassian says only Confluence instances with the “Remote API (XML-RPC & SOAP)” setting turned on are exposed; it has been off by default for new installs since Confluence 3.0 and was removed entirely in Confluence 10.0.

How fast did the Confluence attack move?

Quickly. In the case Rapid7 flagged on November 13, the attackers ran shell commands and scripts pulled from temporary tunneling domains, and a little over three minutes after their first command they were logging in to the MySQL database with Confluence’s own credentials. Their scripts, which Rapid7 believes were likely AI-generated, went through uploaded attachments and Office documents looking for credentials, tokens, VPN configurations and private keys, and even hunted for payment card numbers. They mapped internal services, queried user and group data, and tried a Kerberos credential-harvesting technique called AS-REP roasting. About 12 hours after the first exploit, they signed in to Microsoft 365 as an integration service account using legacy authentication.

Is there a patch for each one?

Not in the usual sense for Axis. Axis 1.x has reached end of life and the flaw remains unpatched; the recommendation is to move to another SOAP engine such as Apache Axis 2/Java, or use an Axis build that includes a fix committed on July 12, 2026. For the others, apply the vendors’ updates and mitigations. Owners of Confluence should confirm the remote API setting is off or the product is current.

What deadline did CISA set?

Federal civilian agencies must fix or mitigate the DSM, REDCap and Axis flaws by November 21, 2026, and the LiteLLM flaw by December 2, 2026. Those deadlines legally apply only to federal agencies, but CISA encourages every organization to prioritize KEV-listed flaws, since they are proven targets.

What this means for your company

The KEV list is one of the most useful free tools a small business has, because it separates “theoretical” from “being used right now.” Start with the devices people forget: a NAS box in a closet that holds your backups or shared files should be on current firmware and should not be reachable from the internet. Ask your software vendors and IT provider whether any of your applications embed Apache Axis or expose old remote APIs, and switch off features you do not use. If your team is experimenting with AI gateways or similar tools, treat them like any other server: patch them, keep admin endpoints private and rotate keys. Finally, the Confluence case shows why legacy authentication to Microsoft 365 and stored service-account passwords are risky; block legacy sign-in methods and give service accounts only the access they need.

If you would like help checking your network storage, business applications and cloud sign-in settings against this week’s list, we can walk through it with you. Reach out to Layer 7 Data Solutions to discuss how we’ve helped companies similar to yours.

About The Author