REDCap unauthenticated RCE exploited in the wild (CVE-2026-90817)
At a glance
| Severity | CRITICAL |
|---|---|
| CVSS | CVE-2026-90817: 9.8 |
| EPSS (30-day exploit probability) | CVE-2026-90817: 25.2% |
| In CISA KEV (exploited) | Yes, due 2026-11-21 |
| Vendor | Vanderbilt University |
| CVE IDs | CVE-2026-90817 |
Primary-source details
- CVE-2026-90817: CVSS 9.8 (Critical), per NVD; fixed in vanderbilt redcap: 16.0.49, 17.3.10, 17.4.4 (NVD); added to the CISA KEV catalog on 2026-11-18, remediation due 2026-11-21.
CISA has added a REDCap vulnerability, CVE-2026-90817, to its Known Exploited Vulnerabilities catalog, which means it is being exploited in the wild. The flaw is an unauthenticated remote code execution bug in Vanderbilt University’s Research Electronic Data Capture (REDCap) platform, rated CVSS 9.8. Federal agencies must remediate by 21 November 2026, according to the CISA KEV catalog.
What happened
CISA listed the REDCap code injection flaw on 18 November 2026. The catalog describes it as a bug that lets an unauthenticated attacker run arbitrary code on the server. The attack works through crafted HTTP requests that reach an unintended controller route from a public survey context, combined with a crafted file-path or stream parameter during data import processing.
Exploitation requires knowledge of a valid public survey hash. That is a real hurdle, but survey links are shared with participants by design, so it should not be treated as a strong barrier.
CISA has not said whether the flaw is used in ransomware campaigns; the catalog lists that as unknown. Details of the exploitation activity, such as who is behind it or how many servers are affected, have not been published.
Am I affected?
NVD says the flaw exists in REDCap 13.3.0 and higher. The affected ranges are:
- 13.3.0 up to, but not including, 16.0.49
- 16.1.0 up to, but not including, 17.3.10
- 17.4.0 up to, but not including, 17.4.4
The fixed releases listed by NVD are 16.0.49, 17.3.10 and 17.4.4. Any instance that exposes public surveys to the internet is the most likely to be reachable.
Technical details
- Weakness: CWE-94 (code injection) and CWE-73 (external control of a file name or path).
- CVSS 3.1: 9.8 Critical, vector AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H. That means network reachable, low complexity, no privileges and no user interaction.
- EPSS: about 25%, the modelled probability of exploitation in the next 30 days.
- KEV: added 18 November 2026, due date 21 November 2026.
Successful exploitation gives the attacker code execution on the REDCap server, with high impact on confidentiality, integrity and availability.
Why this matters
This section is analysis, not reported fact. REDCap is typically used to collect research and clinical study data, so a server compromise can expose participant records and sit next to other sensitive systems. A KEV listing with a three-day federal deadline signals that defenders should treat this as a patch-within-hours item, not a routine maintenance window. The need for a survey hash lowers the risk only slightly when surveys are public.
What to do
- Identify every REDCap instance, including test and departmental servers, and record the version of each.
- Upgrade to 16.0.49, 17.3.10 or 17.4.4, whichever matches your release line.
- If you cannot patch immediately, follow the CISA required action: apply the vendor’s mitigations, or stop using the product if mitigations are unavailable. Consider taking public survey access offline until the upgrade is done.
- Review web server logs for unusual requests to survey routes and data import endpoints, especially requests with unexpected file-path or stream parameters.
- Look for new or modified files in the web root, unexpected child processes of the web server, and outbound connections from the REDCap host.
- CISA also points to its forensics triage requirements. If you find signs of compromise, preserve evidence before rebuilding the server and rotate credentials stored on it.
- Tell your research data owners and privacy office if the server held participant data and you cannot rule out access.
Detection and hunting ideas
No indicators of compromise have been published. Hunt on behaviour instead: web requests to public survey URLs followed shortly by import-related requests from the same address, web server processes spawning shells, and files written by the web server account outside the normal upload directories.
