Key Points
- CISA added four actively exploited vulnerabilities in Synology DSM, LiteLLM, Apache Axis and REDCap to its KEV catalog.
- Three carry CVSS scores of 9.8, and the Apache Axis flaw has no patch because Axis 1 reached end-of-life.
- Federal agencies face fix deadlines of November 21 and December 2, 2026, depending on the product.
The latest:
Four vulnerabilities now under active attack were added Wednesday to the U.S. Cybersecurity and Infrastructure Security Agency’s Known Exploited Vulnerabilities catalog, covering Synology DiskStation Manager, BerriAI LiteLLM, Apache Axis and Vanderbilt University’s REDCap. Three of the four score 9.8 on the CVSS severity scale. The observed campaigns span ransomware, cryptocurrency mining and the theft of AI model access.
Details:
- The Axis flaw: CVE-2026-90436, scored 9.8, is an unsafe reflection bug in Apache Axis that CISA said lets an unauthenticated remote attacker run arbitrary code. It affects Axis 1.x, which is end-of-life and unpatched. CISA recommends migrating to Apache Axis 2/Java or applying a build carrying a July 12, 2026 commit.
- Confluence exposure: Rapid7 said it detected exploitation of the Axis flaw against Atlassian Confluence Data Center, because the platform uses the Axis library in its XML-RPC and SOAP APIs. Atlassian said an instance is only vulnerable with Remote API enabled, a setting off by default since Confluence 3.0 and removed entirely in Confluence 10.0.
- Inside the intrusion: Rapid7 flagged the incident on November 13, 2026, and said attackers pulled shell scripts from temporary Pinggy and Cloudflare Tunnel domains. Within roughly three minutes of the first execution, they were authenticating to MySQL with Confluence’s own database credentials and querying page bodies, application links and attachments.
- What they hunted: According to Rapid7, scripts it assessed as likely AI-generated parsed Office documents in Confluence’s attachment directory, searched for credentials, tokens, VPN configurations and private keys, and hunted payment card numbers using Luhn checks to filter false positives.
- Escalation attempt: Rapid7 said the attackers queried the Atlassian Crowd API for users and group memberships and attempted AS-REP Roasting for Kerberos credential harvesting. Those efforts failed, but the actor obtained a service account credential, queried a production cloud data warehouse, and about 12 hours in signed into Microsoft 365 via legacy authentication.
- The REDCap bug: CVE-2026-90817, scored 9.8, is a code injection flaw allowing unauthenticated remote code execution. CISA said exploitation runs through crafted HTTP requests reaching an unintended controller route from a public survey context, supplying a manipulated file-path parameter during data import, and requires a valid public survey hash.
- Mining payload: Safe Security reported late last month that an unidentified threat actor abused the REDCap flaw to “stage a transient web shell, followed by an in-memory loader that retrieved XMRig disguised as image files, masqueraded the miner as syslog-ng, and established persistence through cron and an added SSH key.”
- AI access theft: The LiteLLM flaw, CVE-2026-40217 at 8.8, allows remote code execution through bytecode rewriting at a guardrails testing endpoint. Google Threat Intelligence Group reported this week that attackers escalated limited inference keys into full administrative credentials in LLMjacking operations against financial services, retail, transportation and technology firms across three regions.
- Ransomware link: CVE-2024-10441, scored 9.8, is an improper output encoding flaw in Synology DSM. Synology disclosed earlier this month that it is being exploited in ransomware attacks on network attached storage devices, with intrusions ending in deployment of C3RB3R ransomware. The exploitation vectors were not specified.
- Remediation clock: CISA set a November 21, 2026 deadline for federal agencies to patch or mitigate the DSM, REDCap and Axis flaws, and December 2, 2026 for the LiteLLM flaw. KEV listings bind federal civilian agencies, though private operators widely treat the catalog as a priority patching list.
Between the lines:
The four entries point at different attacker economics rather than one campaign: NAS ransomware monetizes extortion, the XMRig chain monetizes stolen compute, and LLMjacking monetizes stolen model access. The Axis case is the structural problem, since an end-of-life library embedded in Confluence leaves defenders with migration rather than a patch.
What’s next
Watch whether exploitation of the Axis flaw spreads beyond Confluence given no vendor patch exists, whether Google Threat Intelligence Group reports further LLMjacking victims, and whether agencies meet the November 21 and December 2 deadlines.