CISA Added Synology DSM, Apache Axis and LiteLLM to Its Exploited List. Synology and Axis Are Due Saturday, LiteLLM December 2. Here Is the Version Check for Each

If you run any of these three, act this week:
- Synology DSM or BeeStation (CVE-2024-10441): update to the fixed build in the table below by Saturday, November 21. Better, install the newest release for your model.
- Anything built on Apache Axis 1.x (CVE-2026-90436): there is no fixed release and there will not be one. By Saturday, either migrate off Axis 1, run a build with Apache's patch applied, or take the SOAP endpoint off the internet.
- LiteLLM proxy (CVE-2026-40217): upgrade to 1.83.11 or later by Wednesday, December 2. Until then, block POST requests to /guardrails/test_custom_code at your reverse proxy.
CISA added all three to its Known Exploited Vulnerabilities catalog on Wednesday, November 18, along with REDCap CVE-2026-90817, which we covered separately: REDCap CVE-2026-90817 is exploited and in KEV. The deadlines are for US federal agencies, but they are a reasonable clock for everyone. November 21 is a Saturday, so for most teams the real deadline is Friday.
The Hacker News headlined the batch "CISA Warns of 4 Exploited Flaws Linked to Ransomware, Cryptomining, and LLMjacking Attacks." CISA itself marks known ransomware use as "Unknown" for all four. Below, for each CVE, we say which of those three labels traces to a source we read, and which comes only from press reports.
The dates, from CISA's own file
From the CISA KEV JSON feed, read November 19:
- CVE-2024-10441, Synology DiskStation Manager. Added November 18, due November 21. Known ransomware use: Unknown. Forensic triage: Yes.
- CVE-2026-90436, Apache Axis. Added November 18, due November 21. Known ransomware use: Unknown. Forensic triage: Yes. CISA notes it "could affect an open-source component, third-party library, protocol, or proprietary implementation that could be used by different products," and that the product could be end of life.
- CVE-2026-40217, BerriAI LiteLLM. Added November 18, due December 2. Known ransomware use: Unknown. Forensic triage: No.
For all three, the required action is to apply vendor mitigations under BOD 26-04, or stop using the product if mitigations are unavailable.
Synology DSM, [CVE-2024-10441](https://analytics.dugganusa.com/api/v1/dredd/kev-gap?cve=CVE-2024-10441)
What it is. Synology's advisory Synology-SA-24:20 (published November 5, 2024, last updated March 28, 2025, status Resolved) rates it Critical, CVSS 9.8: an "improper encoding or escaping of output vulnerability in the system plugin daemon" that "allows remote attackers to execute arbitrary code via unspecified vectors." No credentials are needed. Synology lists no workaround, so updating is the only fix.
The ransomware link. On October 29, Synology said attackers were exploiting a vulnerability it had already patched, and pointed to SA-24:20. That notice did not name a CVE. CISA has now listed CVE-2024-10441 as exploited, which settles the "is it being used" question for this CVE. The ransomware name (C3RB3R, files ending in .L0CK3D) comes from press reports, not from Synology. We covered the outbreak on November 2: ransomware is locking Synology NAS boxes through a bug patched almost two years ago. In that post we called CVE-2024-10441 our inference for the entry point; CISA's listing now backs the "exploited" part, though CISA still does not say it is the ransomware's way in.
Am I affected? In DSM, open Control Panel, then Info Center, then General, and read "DSM version." You are on a fixed build if it is at or above:
- DSM 7.2.2: 7.2.2-72806-1
- DSM 7.2.1: 7.2.1-69057-6
- DSM 7.2: 7.2-64570-4
- DSM 7.1: 7.1.1-42962-7
- DSM 6.2: 6.2.4-25556-8 (Synology rates the 6.2 line Moderate in this advisory, not Critical)
- BeeStation OS (BSM): 1.1-65374, per the CVE text in the same advisory
- DSM UC: the advisory's revision history says a DSM UC 3.1 update became available November 14, 2024, but does not give the build number
If your box is reachable from the internet through port forwarding or QuickConnect, move management behind a VPN. That is our advice, not Synology's. And check that you have a backup the NAS itself cannot overwrite.
Apache Axis 1.x, [CVE-2026-90436](https://analytics.dugganusa.com/api/v1/dredd/kev-gap?cve=CVE-2026-90436)
What it is. Apache's announcement on its Axis mailing list (November 6, 2026) titles it "Apache Axis 1.x (EOL): unauthenticated remote class instantiation via SOAP fault exceptionName," severity critical. When an incoming SOAP fault names an exception class, Axis loads that class, and can construct it, before checking that it is a real fault. "With a suitable class on the application's classpath this leads to remote code execution." Apache adds that "authentication handlers configured on a service do not protect against it," because the fault is parsed before any per-service handler runs. All Axis 1.x versions through 1.4 are affected. The CVE record is marked "UNSUPPORTED WHEN ASSIGNED."
There is no fixed release. Apache's words: "The Apache Axis project does not expect to create an Axis 1.x release fixing this problem." It recommends migrating to another SOAP engine such as Apache Axis2/Java, or using a build of Axis with the patch from commit 41f3b4e8 in the axis-axis1-java repository applied. That commit is dated July 12, 2026. It makes Axis load the named class without running its static initializer and refuse any class that is not an AxisFault.
The cryptomining link. This one traces to a source we read. Huntback's November 12 research documented attackers sending a crafted SOAP fault to N-able N-central at /dms/services/ServerUI, naming a class to instantiate, then installing the XMRig Monero miner on 21 servers it could see mining. That is the 4EverAble2Shell chain; the indicators Huntback published and the file paths to hunt for are in our post N-central 4EverAble2Shell is now exploited, and the background is in new public exploits claim to beat N-central's HF2 fix. The Hacker News also reports that Rapid7 saw exploitation against Atlassian Confluence Data Center for data theft and reconnaissance, and quotes Atlassian that a Confluence instance must have "Remote API (XML-RPC & SOAP)" enabled to be vulnerable. We did not read Rapid7's or Atlassian's notices, so treat the Confluence part as press reports.
Am I affected? Axis 1.x usually arrives inside another product, so look for the library, not a product name:
- Search your servers and application folders for a file named axis.jar or axis-1.4.jar, typically under a WEB-INF/lib folder. Open its META-INF/MANIFEST.MF; any 1.x version is affected.
- In build files, look for the Maven coordinates axis:axis or org.apache.axis:axis. Every published version is 1.4 or earlier.
- In web.xml, look for the servlet class org.apache.axis.transport.http.AxisServlet, and in access logs for request paths containing /services/ (the Axis default is /axis/services/; N-central uses /dms/services/).
- Axis2 (Maven group org.apache.axis2) is a separate code base and is not named in this advisory.
If you find Axis 1.x in a commercial product, the fix has to come from that vendor; ask them which build carries the patch. Until then, the SOAP endpoint should not be reachable from the internet.
LiteLLM, [CVE-2026-40217](https://analytics.dugganusa.com/api/v1/dredd/kev-gap?cve=CVE-2026-40217)
What it is. BerriAI's GitHub advisory GHSA-wxxx-gvqv-xp7p, "Sandbox escape in custom-code guardrail" (published May 7, 2026, rated medium by BerriAI). The POST /guardrails/test_custom_code endpoint runs user-supplied Python inside a hand-rolled sandbox, which "can be escaped using bytecode-level techniques, allowing arbitrary code execution in the proxy process — which runs as root in the default Docker image." The fix replaced the sandbox with RestrictedPython.
The LLMjacking link. This one traces to a source we read. Google Threat Intelligence Group's November 16 report lists it in a table titled "Common TTPs observed in LLMJacking intrusions": "We also observed a threat actor exploit CVE-2026-40217, a sandbox escape in LiteLLM that allows an attacker to elevate limited-access inference keys ('Virtual Keys') into full administrative credentials." Background on why gateway keys are the prize: AI agents stole AI keys from exposed LiteLLM, Dify and One API servers and Google put numbers on LLMjacking.
Two sources that do not line up. BerriAI says "reaching the endpoint requires a proxy-admin credential in default configurations." Google says the flaw let an attacker climb from a limited virtual key to admin. Google gives no path and does not say whether another flaw or a non-default setup was involved. We cannot reconcile the two from what was published. Patch regardless: either way it ends in code running as root on the box that holds every model-provider key you have. Also, our November 4 post described CVE-2026-40217 as a flaw "in the key regeneration path." BerriAI's advisory describes the custom-code guardrail sandbox, and that is the description to trust.
Am I affected? Run pip show litellm in the proxy's environment, or read the image tag of your LiteLLM container.
- Affected: 1.81.8 up to and including 1.83.9 (the advisory's version range, and OSV's).
- The advisory's text says "Fixed in 1.83.11," while its version range lists 1.83.10 as patched and links the v1.83.10-stable release. Go to 1.83.11 or later and you are clear under both readings.
- When you upgrade, do not land on 1.107.0, the release that was briefly backdoored on October 21: LiteLLM was poisoned again, through a different door.
- If you cannot upgrade today, BerriAI's workaround is to block POST /guardrails/test_custom_code at your reverse proxy or API gateway. Then search your proxy access logs for that path; any request to it you did not make yourself is worth a forensic look.
- Because Google ties this to key theft, rotate the upstream provider keys the proxy holds if your instance was reachable from the internet, and check provider billing for spend you do not recognize.
What our package check says. We ran our check-package tool on litellm 1.83.9 (vulnerable), 1.83.10 and 1.83.11 (fixed). It returned the same "review" verdict for all three, because LiteLLM is on our deny-list for an earlier malicious release (OSV MAL-2026-2144), and it reported each version as outside that record's affected range. It did not report CVE-2026-40217 against 1.83.9. So today it does not tell a vulnerable LiteLLM from a fixed one for this CVE; use the version list above.
What is not known
- Indicators. None of the sources we read for Synology or LiteLLM publish addresses, domains or hashes for this exploitation. We are adding nothing to our feed for them. The Axis indicators that exist are the N-central ones, in the post linked above.
- Whether CVE-2024-10441 is the ransomware's entry point. CISA says it is exploited; nobody we read ties the C3RB3R/L0CK3D infections to it by name.
- Which other products bundle Axis 1.x and are being hit. Only N-central is documented in a source we read.
- How the LiteLLM attacker reached an admin-only endpoint, per the conflict above.
- Our own edge. Our edge-shield protects our web properties, not your NAS, your Java app server or your AI gateway. We scanned our edge blocks (550,543 since October 20), edge canaries (94,728) and origin honeypot captures (695) for /dms/services/ServerUI, /guardrails/test_custom_code and the Axis public proof-of-concept path, and found zero matches. That says nobody probed our sites for those paths. It says nothing about whether your systems are being probed. Synology's flaw has no published HTTP path to look for.
How we know
We read CISA's KEV JSON feed for every date and field above. We read Synology's SA-24:20 advisory, Apache's announcement on the Axis mailing list, the CVE-2026-90436 record, the Axis patch commit, BerriAI's GitHub advisory and its OSV record, and Google's November 16 report directly. The Huntback findings come from our November 12 post, where we read Huntback's research. The Hacker News article was our pointer to the batch and is our only source for the Rapid7/Confluence detail. We have not tested any exploit.
Sources
- CISA, Known Exploited Vulnerabilities catalog (JSON feed), entries for CVE-2024-10441, CVE-2026-90436, CVE-2026-40217 and CVE-2026-90817, added November 18, 2026
- Synology, Synology-SA-24:20 DSM (PWN2OWN 2024): synology.com/en-global/security/advisory/Synology_SA_24_20
- Apache Axis, "CVE-2026-90436: Apache Axis 1.x (EOL): unauthenticated remote class instantiation via SOAP fault exceptionName," java-user mailing list, November 6, 2026: lists.apache.org/thread/8slf0j9gz35z56t0glt5h0ncwlbn4bdz
- Apache, axis-axis1-java commit 41f3b4e8d04e310ca81f86b92630ab2c11d608bd, July 12, 2026
- BerriAI, GHSA-wxxx-gvqv-xp7p, "Sandbox escape in custom-code guardrail": github.com/BerriAI/litellm/security/advisories/GHSA-wxxx-gvqv-xp7p
- Google Threat Intelligence Group, "Agentic threat activity crosses a significant threshold," November 16, 2026
- Huntback, N-central 4EverAble2Shell miners on live servers, November 12, 2026 (via our post of the same day)
- The Hacker News, "CISA Warns of 4 Exploited Flaws Linked to Ransomware, Cryptomining, and LLMjacking Attacks," November 19, 2026
Am I affected? The short version
- Synology: Control Panel, Info Center, General, "DSM version," against the table above.
- Axis: search for axis.jar or axis-1.4.jar and the AxisServlet class; any hit is affected.
- LiteLLM: pip show litellm; anything from 1.81.8 through 1.83.9 is affected, move to 1.83.11 or later.
Read next: REDCap CVE-2026-90817, the fourth entry in this batch, also due Saturday.
Every indicator in this post is in the feed. Free.
1.58M+ IOCs, STIX 2.1 / TAXII, 88% novel vs ThreatFox, exploited-CVE leads ahead of CISA. No credit card — a free API key in 30 seconds, and you can audit every claim above against the live endpoints.